Treatment response, drug survival, and predictors thereof in 764 patients with psoriatic arthritis treated with anti–tumor necrosis factor α therapy: Results from the nationwide Danish DANBIO registry

Authors

  • Bente Glintborg,

    Corresponding author
    1. Gentofte University Hospital, Hellerup, Denmark
    • Gentofte University Hospital, Niels Andersensvej 65, 2900 Hellerup, Denmark
    Search for more papers by this author
  • Mikkel Østergaard,

    1. Copenhagen University Hospital at Glostrup, Denmark
    Search for more papers by this author
    • Dr. Østergaard has received consulting fees, speaking fees, and/or research grants from Abbott, Amgen, Bristol-Meyers Squibb, Centocor, Genmab, GlaxoSmithKline, Novo Nordisk, Wyeth/Pfizer, Roche, Schering-Plough/MSD, and UCB (less than $10,000 each).

  • Lene Dreyer,

    1. Copenhagen University Hospital, Rigshospitalet, Copenhagen, Denmark
    Search for more papers by this author
  • Niels Steen Krogh,

    1. Zitelab Aps, Copenhagen, Denmark
    Search for more papers by this author
  • Ulrik Tarp,

    1. Århus University Hospital, Århus, Denmark
    Search for more papers by this author
  • Michael Sejer Hansen,

    1. Gentofte University Hospital, Hellerup, Denmark
    Search for more papers by this author
    • Dr. Hansen has received consulting fees, speaking fees, and/or research grants from Abbott, Roche, and Schering-Plough/MSD (less than $10,000 each).

  • Signe Rifbjerg-Madsen,

    1. Frederiksberg Hospital, Frederiksberg, Denmark
    Search for more papers by this author
  • Tove Lorenzen,

    1. Vejle Hospital, Vejle, Denmark
    Search for more papers by this author
    • Dr. Lorenzen serves on the Danish Advisory Board of Abatacept, Bristol-Myers Squibb and is a member of the Abbott Scientific Committee.

  • Merete Lund Hetland

    1. Copenhagen University Hospital at Glostrup, and DANBIO Registry, Copenhagen, Denmark
    Search for more papers by this author
    • Dr. Hetland has received consulting fees, speaking fees, and/or research grants from Abbott, Centocor, Roche, Schering-Plough/MSD, UCB-Nordic, and Wyeth/Pfizer (less than $10,000 each) and, on behalf of DANBIO, she has received grants from Abbott, Bristol-Myers Squibb, Roche, Schering-Plough/MSD, UCB-Nordic, and Wyeth/Pfizer.


Abstract

Objective

To investigate disease activity, treatment response, and drug survival, and predictors thereof, among Danish patients with psoriatic arthritis (PsA) receiving their first treatment series with a tumor necrosis factor α (TNFα) inhibitor.

Methods

Patients with PsA were identified from a prospective nationwide rheumatologic database, the Danish biologics registry DANBIO, using data registered from 2000–2009. Information was obtained on the patients' clinical response to anti-TNFα treatment (defined as achievement of the American College of Rheumatology 20% [ACR20], ACR50, and ACR70 improvement criteria or a European League Against Rheumatism [EULAR] good response at least once during the first 6 months of treatment) and duration and rate of drug adherence (referred to as drug survival), as well as predictors thereof.

Results

Of 764 patients with PsA, 320 received adalimumab, 260 infliximab, and 184 etanercept. Median drug survival was 2.9 years, and 1-year and 2-year drug survival rates were 70% and 57%, respectively. Clinical parameters that showed improvement over 6 months were the C-reactive protein (CRP) level, Health Assessment Questionnaire score, and 28-joint Disease Activity Score. Male sex, CRP level >10 mg/liter, concomitant methotrexate use, and low patient health visual analog scale score at baseline were associated with longer drug survival. Improvement was achieved by 59%, 45%, 24%, and 54% of patients according to the ACR20, ACR50, ACR70 response criteria and EULAR good response, respectively. A CRP level >10 mg/liter was predictive of the improvement responses (odds ratio [OR] 2.6 for ACR20, OR 3.0 for ACR50, OR 3.6 for ACR70, and OR 2.2 for EULAR good response).

Conclusion

In these patients with PsA treated with their first TNFα inhibitor in clinical practice, high drug adherence and responder rates were observed. Moreover, increased levels of CRP at baseline were associated with both good treatment responses and continued treatment, which may be of clinical value in selecting the patients most likely to benefit from treatment with TNFα inhibitors.

The prevalence of psoriatic arthritis (PsA) among Caucasians is 0.1–0.2%, with an estimated incidence of 6 per 100,000 person-years (1). PsA covers a wide spectrum of disease manifestations, including peripheral and axial arthritis, enthesitis, and dactylitis (2, 3).

The wide range of disease manifestations in PsA warrants individualized treatment and presents a challenge to the treating physician (4). Similar to rheumatoid arthritis (RA), disease-modifying antirheumatic drugs (DMARDs) are often the first line of treatment in PsA. However, the experience with DMARDs has mainly been obtained in RA patients, and only a few randomized clinical trials have been performed in patients with PsA (4). In the past decade, several placebo-controlled clinical trials have described the effect of tumor necrosis factor α (TNFα)–inhibiting therapy in PsA (2, 4, 5). Until recently, 3 TNFα inhibitors, infliximab, adalimumab, and etanercept, were available for the treatment of PsA (4, 6).

National registries that include followup data on patients treated with biologic drugs have been established in several countries (5, 7–9). These real-life data allow for the investigation of long-term drug effects among a heterogeneous group of patients, including elderly patients who may experience comorbid diseases. Such data provide a valuable supplement to results from randomized clinical trials, which often include a limited number of patients who fulfilled a strict set of inclusion criteria and who were studied for a limited time span (5, 10, 11).

The rheumatologic database of the Danish biologics registry DANBIO currently includes up to 8 years of followup data. It is mandatory for the treating clinicians to prospectively report treatment and disease activity among patients being treated with biologic agents for a rheumatic disease (12, 13). Our aims, based on the DANBIO data, were to report drug efficacy and drug survival (defined as duration and rate of adherence to anti-TNFα treatment), as well as to identify predictors thereof, among Danish patients with PsA who were receiving their first treatment course with a TNFα inhibitor in routine care.

PATIENTS AND METHODS

Identification of patients.

DANBIO is a nationwide Danish rheumatologic registry that was initiated in 2000 and covers ∼90% of patients treated with a biologic drug in routine clinical care (12, 13). Each year, all departments of rheumatology in Denmark (25 in total) are asked to report the personal identification code of all patients receiving biologic treatment. When the codes in the registry were compared with these reports, coverage was calculated to be 88% in 2008 and 93% in 2009. By November 1, 2009, 1,237 patients with PsA, diagnosed according to specialists in rheumatology, had been registered in DANBIO. Of these patients, 432 were not included in the present study, because they were only treated with DMARDs and had never received biologic treatment. Another 41 patients were treated with biologic agents but participated in clinical studies, and therefore were also excluded. Thus, 764 patients were included in the present study (Table 1). Patients treated by dermatologists are not included in the database.

Table 1. Characteristics of the 764 patients with psoriatic arthritis*
  • *

    IQR = interquartile range; MTX = methotrexate; TNFα = tumor necrosis factor α.

  • Some patients previously received >1 disease-modifying antirheumatic drug (DMARD), and therefore the percentages do not add up to 100%.

Demographic variables at baseline
 Women, no. (%)396 (52)
 Age, median (IQR) years47 (38–56)
 Disease duration, median (IQR) years5 (2–11)
 Symptom duration, median (IQR) years7 (4–14)
 MTX use, no. (%)410 (54)
Previous DMARDs, no. (%)
 Sulfasalazine391 (51)
 MTX288 (38)
 Glucocorticoids147 (19)
 Cyclosporine67 (9)
 Azathioprine51 (7)
 Gold47 (6)
 Leflunomide38 (5)
 Not stated97 (13)
TNFα inhibitor, no. (%)
 Adalimumab320 (42)
 Etanercept184 (24)
 Infliximab260 (34)
Year of treatment initiation, no. (%)
 20002 (0.3)
 20017 (1)
 20028 (1)
 200341 (5)
 200465 (8)
 2005113 (15)
 2006118 (16)
 2007128 (17)
 2008161 (21)
 2009 (January–November)121 (16)
Reason for drug discontinuation, no. (%)
 Lack of efficacy175 (23)
 Adverse effects95 (12)
 Disease remission9 (1)
 Psoriatic skin flare7 (1)
 Planning pregnancy6 (1)
 Cancer (suspected/verified)5 (1)
 Lost to followup4 (1)
 Other reasons20 (3)
 Not stated15 (2)
 Total discontinuations336 (44)

In Denmark, biologic treatments of rheumatic diseases can only be prescribed and administered at hospital departments of rheumatology and not by private practitioners. The drugs are fully reimbursed by the health care system and are of no cost to the individual patient. Therefore, the selection of patients for treatment is not influenced by factors such as social status. According to national clinical guidelines, patients who are considered for treatment should have continuously active disease and must have experienced treatment failure with at least 1 DMARD. The decision to start and stop treatment is made locally by a rheumatologist.

Baseline demographic and clinical variables included in the registry are age, sex, disease duration, and previous or current treatment with methotrexate (MTX) or other DMARDs. In addition, disease activity parameters are prospectively reported to DANBIO through an online system (available at http://www.danbio-online.dk) (14). Data collection occurs, at a minimum, biannually.

The DANBIO database has been approved by the Danish Board of Health and the Danish Data Registry. The registration of data on patients treated with biologic agents does not require patient consent. Publication of data does not require approval by the Registry Ethics Committee.

Measures of disease activity and clinical response.

Levels of disease activity were monitored with the use of Health Assessment Questionnaire (HAQ) scores (15), the 28-joint Disease Activity Score (DAS28) (16), and visual analog scale (VAS) scores for pain, patient's global assessment of health, and fatigue. In cases of clinical signs and/or symptoms of axial involvement, the treating rheumatologist was encouraged to additionally monitor axial disease activity, using the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), Functional Index (BASFI), and Metrology Index (BASMI) (introduced in 2003) (17–19). In addition, disease activity and outcomes were evaluated using the C-reactive protein (CRP) level and the swollen and tender joint counts. All of these measures were determined at 0, 2, and 6 weeks, 6 months, and 1, 2, 3, 4, and 5 years after initiation of the anti-TNFα therapy (Table 2).

Table 2. Disease activity in patients with psoriatic arthritis at baseline and during followup*
 Baseline2 weeks6 weeks6 months1 year2 years3 years4 years5 years
  • *

    Except where indicated otherwise, values are the median (interquartile range). The baseline visit was the time point at which the patient received his or her first dose of tumor necrosis factor α inhibitor. Outcome data were reported according to the registrations in the DANBIO registry at the following time points: 2 weeks of treatment (time interval 1–4 weeks), 6 weeks (time interval 5–9 weeks), 6 months (time interval 18–32 weeks), 1 year (time interval 46–64 weeks), 2 years (time interval 91–117 weeks), 3 years (time interval 143–182 weeks), 4 years (time interval 183–233 weeks), 5 years (time interval 234–285 weeks), and 6 years (time interval 286–338 weeks). If more DANBIO registrations occurred within a given time interval, the one closest to the given time point was selected. If a patient had no registrations within a given time interval, data were registered as missing for the given time point. CRP = C-reactive protein; HAQ = Health Assessment Questionnaire; DAS28 = Disease Activity Score in 28 joints; VAS = visual analog scale; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; BASFI = Bath Ankylosing Spondylitis Functional Index; BASMI = Bath Ankylosing Spondylitis Metrology Index.

  • P < 0.001 by paired comparison of followup data at all time points versus baseline.

  • P < 0.05 by paired comparison of followup data at all time points versus baseline.

  • §

    P < 0.05 at this time point versus baseline.

  • Among patients with incomplete followup data, the followup time was estimated according to the latest visit registered in the DANBIO.

CRP, mg/liter10 (4–22)5 (1–9)5 (1–10)4 (1–9)5 (1–10)5 (1–10)4 (2–9)4 (1–9)4 (1–8)
No. of swollen joints3 (1–7)1 (0–4)0 (0–1)0 (0–1)0 (0–1)0 (0–1)0 (0–1)0 (0–1)0 (0–1)
No. of tender joints7 (3–13)3 (1–8)2 (0–6)1 (0–5)1 (0–3)0 (0–2)0 (0–3)0 (0–2)0 (0–2)
HAQ score1.0 (0.6–1.5)0.75 (0.25–1.25)0.6 (0.1–1.1)0.6 (0.0–1.0)0.4 (0.0–1.0)0.4 (0.0–1.0)0.3 (0.0–1.0)0.3 (0.0–0.9)0.5 (0.0–1.0)
DAS284.8 (3.9–5.5)3.4 (2.6–4.3)3.0 (2.1–4.0)2.8 (1.9–3.9)2.6 (1.8–3.6)2.3 (1.7–3.1)2.1 (1.7–3.2)2.0 (1.5–2.9)2.3 (1.7–2.9)
VAS pain score, mm63 (45–75)36 (20–60)27 (13–52)21 (10–47)20 (8–48)20 (8–38)21 (8–40)24 (10–44)22 (5–46)
VAS fatigue score, mm65 (49–79)51 (29–75)44 (22–71)35 (14–59)37 (10–62)31 (10–52)19 (8–53)29 (7–67)22 (12–35)
VAS global score, mm69 (50–81)46 (21–68)31 (15–58)27 (10–52)23 (9–51)21 (10–45)21 (8–48)20 (7–43)22 (9–39)
BASDAI, mm62 (46–75)49 (28–68)34 (22–52)25 (11–47)25 (9–49)17 (10–47)16 (7–47)16 (6–26)14 (1–42)
BASFI, mm52 (35–68)47 (26–62)38 (21–55)26 (6–47)27 (7–46)18 (6–57)22 (5–57)23 (5–48)23 (6–64)
BASMI, mm20 (10–40)20 (10–30)§10 (10–30)§10 (0–20)§20 (0–30)§10 (0–30)20 (10–50)15 (0–40)40 (10–50)
No. of patients treated76474973261543229318113068
No. of patients with visit registered65827536640631822912710445

The clinical response to anti-TNFα therapy was evaluated as achievement of the American College of Rheumatology 20% (ACR20), ACR50, and ACR70 improvement criteria (20) or a European League Against Rheumatism (EULAR) good response (21). Arbitrarily, we classified patients as responders if they achieved a clinical response (determined as yes versus no) at least once during the first 6 months of treatment. Complete baseline and outcome data for the calculations of the ACR and EULAR responses were available from 426 patients (56%) and 483 patients (63%), respectively.

Treatment duration.

Duration of drug survival was calculated as the number of days during which individual patients continued their treatment with the first TNFα inhibitor. The start date was the date at which the first dose was administered, and the stop date was the date of the first missed dose. Temporary treatment interruptions (e.g., due to infections or surgery) of <3 months' duration were allowed. All observations were censored at November 1, 2009. Reasons for drug discontinuation were registered.

Queries were sent to the hospitals with regard to 133 patients for whom no followup data were available since July 1, 2009 and for whom no treatment stop date was registered. Subsequently, data on treatment duration were complete for 741 (97%) of the 764 patients. For the remaining 23 patients whose followup data were incomplete, data were censored according to the last visit registered in DANBIO.

Statistical analysis.

Analyses of the data were performed using SPSS software (version 16.0) and SAS software (version 9.0). Demographic and descriptive data are expressed as the median and interquartile range (IQR). Groups were compared by nonparametric testing (for unpaired data, chi-square and Mann-Whitney tests; for paired data, Wilcoxon's signed rank test). Efficacy was analyzed as described above, per protocol. In all statistical tests, P values less than 0.05 were considered statistically significant.

Kaplan-Meier plots, log rank tests, and multivariate Cox regression analyses were used for drug survival analyses. Logistic regression analysis was used for the identification of factors associated with clinical response. In order to visualize stratified Kaplan-Meier drug survival curves, baseline VAS scores for global health status were converted into quartiles, and baseline CRP levels were converted into a binary variable (below versus above 10 mg/liter, according to the detection limit of the CRP level in most centers). For the analysis of time to discontinuation of treatment due to adverse events, discontinuations due to ineffectiveness were treated as censored observations. Similarly, discontinuations due to adverse events were handled as censored observations in the analysis of time to discontinuation due to ineffectiveness.

In the Cox and logistic regression analyses, sex, type of TNFα inhibitor, baseline CRP level (below or equal to 10 mg/liter versus above 10 mg/liter), and baseline MTX use (yes versus no) were included as categorical variables, whereas patient age, disease duration, swollen and tender joint counts, VAS scores, HAQ scores, and DAS28 scores were continuous variables. The variables with least significance were excluded stepwise (backward selection), leaving only statistically significant variables in the model. The Bath Ankylosing Spondylitis data were excluded from the analysis, because only a limited number of patients had data on axial disease at baseline (for the BASDAI n = 159, for the BASFI n = 157, and for the BASMI n = 122). Baseline MTX use was defined as any concurrent use of MTX at baseline, irrespective of dose, and did not include former or later use of the drug.

In the logistic regression analysis, all interactions involving sex, patient age, disease duration, MTX use, drug type, baseline HAQ scores, and DAS28 scores were tested. Each interaction pair was included in the overall statistical model, and was thereafter excluded in the backward selection process if the interaction was statistically nonsignificant. We found no statistically significant interactions.

RESULTS

Patient characteristics.

The number of patients with PsA who initiated their first treatment series with a TNFα inhibitor increased over the years of the registry. The majority of patients received adalimumab (42%) (Table 1). A total of 336 patients (44%) had withdrawn from TNFα inhibitor treatment by November 1, 2009. As shown in Table 1, the most prevalent reasons for drug discontinuation included lack of efficacy (in 175 of 336 patients; 52% of withdrawals) and adverse effects (in 95 of 336 patients; 28% of withdrawals). Adverse effects were infections in 33 patients, allergic reactions in 5 patients, infliximab infusion reactions in 5 patients, skin rash in 12 patients, nausea/fatigue in 8 patients, neutropenia in 3 patients, elevated liver enzyme levels in 2 patients, chest pain in 2 patients, polyneuropathy in 2 patients, and other/unspecified effects in 23 patients. One of the 5 patients who experienced infliximab infusion reactions was receiving concomitant MTX at the time of the reaction.

At baseline, 410 patients (54%) were receiving concomitant MTX. Baseline MTX use was more prevalent among patients receiving infliximab (70%) compared with those receiving adalimumab (49%) or etanercept (39%) (P < 0.001). After 3 months, 329 (80%) of these patients continued to receive concomitant MTX, 44 patients (11%) had stopped receiving MTX but continued to receive TNFα inhibitor treatment, and 37 patients (9%) had stopped receiving the TNFα inhibitor.

When we compared the baseline data between women and men, we found that women had significantly higher HAQ scores (median 1.13 versus 0.88), VAS global health scores (median 72 mm versus 66 mm), VAS fatigue scores (median 70 mm versus 62 mm), VAS pain scores (median 65 mm versus 62 mm), and tender joint counts (median 7 versus 6), whereas women had lower swollen joint counts (median 3 versus 4) (all P < 0.05 by Mann-Whitney test). Age (median 48 years versus 46 years), disease duration (median 4 years versus 6 years), DAS28 scores (median 4.8 for both), and baseline CRP levels (median 10 mg/liter for both) were similar between women and men (P > 0.05).

Therapeutic effect.

All outcome parameters improved during followup (all P < 0.05 versus baseline, by Wilcoxon's signed rank test) (Table 2). Among the 483 patients with available EULAR response data, 259 patients (54%) achieved a EULAR good response at least once during the first 6 months of treatment, 131 (27%) achieved a moderate response, and 93 (19%) had no response. ACR improvement responses were available for 426 patients, among whom 253 (59%) achieved a response at the ACR20 level, 190 (45%) achieved ACR50, and 104 (24%) achieved ACR70.

Drug survival.

The total treatment period was 2,135 person-years. The median duration of drug survival was 2.9 years, and 1- and 2-year drug survival rates were 70% and 57%, respectively.

The crude retention rates were similar among patients receiving infliximab, those receiving adalimumab, and those receiving etanercept (P > 0.05). As shown in Figure 1, male sex (P < 0.001 by log rank test), a CRP level >10 mg/liter at baseline (P = 0.006), and a low VAS score for global health at baseline (P = 0.005) were associated with improved drug survival. Concomitant use of MTX at baseline did not affect drug survival (P > 0.05).

Figure 1.

Crude drug survival curves of tumor necrosis factor α (TNFα) inhibitor use in patients with psoriatic arthritis, stratified according to sex, baseline C-reactive protein (CRP) level, and quartiles of visual analog scale (VAS) scores for patient's global assessment of health. The x-axis shows treatment duration (in years), while the y-axis shows the cumulative survival rate. P values were determined by log rank test.

Baseline disease parameters and patient characteristics were included in a Cox regression analysis in order to identify the baseline factors associated with subsequent discontinuation of TNFα inhibitor treatment. In all patients, the VAS pain scores correlated strongly with VAS global health scores (Spearman's rho = 0.81, P < 0.001). Therefore, the VAS pain score was excluded from the multivariate Cox regression analysis. In the final model, female sex, a high VAS global health score at baseline, a low CRP level at baseline, and lack of concomitant MTX use were associated with shorter drug survival, whereas patient age, type of biologic drug, DAS28 score, HAQ score, tender and swollen joint counts, and VAS fatigue score were not associated with drug survival (Table 3).

Table 3. Baseline statistical predictors of discontinuing treatment with tumor necrosis factor α inhibitors, as determined by multivariate Cox regression analysis*
 Unadjusted analysisFinal model after backward selection
HR (95% CI)PHR (95% CI)P
  • *

    Values are the hazard ratio (HR) with 95% confidence interval (95% CI) determined in 764 patients with psoriatic arthritis. VAS = visual analog scale; HAQ = Health Assessment Questionnaire; DAS28 = Disease Activity Score in 28 joints; CRP = C-reactive protein; MTX = methotrexate.

Sex
 Male1 1 
 Female1.65 (1.33–2.05)<0.0011.42 (1.11–1.80)0.005
No. of swollen joints0.99 (0.97–1.02)0.59 
No. of tender joints1.03 (1.02–1.05)<0.001 
VAS fatigue score1.03 (0.93–1.13)0.62  
VAS pain score1.07 (1.01–1.13)0.012  
VAS global score1.10 (1.04–1.16)0.0011.10 (1.04–1.17)0.001
HAQ score1.32 (1.10–1.58)0.003  
DAS281.09 (0.99–1.21)0.093  
CRP    
 ≤10 mg/liter1.39 (1.10–1.75) 1.40 (1.09–1.78)
 >10 mg/liter10.00610.008
MTX use    
 Yes1 1 
 No1.21 (0.98–1.50)0.0821.37 (1.07–1.75)0.013
Biologic drug
 Infliximab1 
 Adalimumab0.73 (0.58–0.95)0.020  
 Etanercept0.78 (0.60–1.03)0.084  
Age1.00 (0.99–1.01)0.61  

In a stratified multivariate Cox regression analysis that included only adverse effects as the event causing drug termination, female sex (hazard ratio [HR] 1.8, 95% confidence interval [95% CI] 1.3–1.9; P = 0.01), use of infliximab (HR versus adalimumab 0.49, 95% CI 0.29–0.84 [P = 0.01]; HR versus etanercept 0.46, 95% CI 0.25–0.85 [P = 0.01]), lack of concomitant use of MTX (HR 1.67, 95% CI 1.02–2.70; P = 0.04), and a high number of tender joints (HR 1.03/joint, 95% CI 1.01–1.06; P = 0.02) at baseline were statistically significant predictors of shorter drug survival. Similarly, including only lack of efficacy as the event causing drug termination, a CRP level ≤10 mg/liter (HR 2.11, 95% CI 1.49–3.00; P < 0.001) and a higher baseline VAS score for global health (HR 1.18/cm, 95% CI 1.09–1.29; P < 0.001) were statistically significant factors associated with shorter drug survival.

Prediction of clinical response.

In a multiple logistic regression analysis (backward stepwise selection) with EULAR good response as the dependent variable, a CRP level of >10 mg/liter (odds ratio [OR] 2.2 for high versus low, 95% CI 1.5–3.2; P < 0.001), male sex (OR 1.5 for male versus female, 95% CI 1.0–2.2; P = 0.04), and younger age (OR 0.98/year increase, 95% CI 0.97–1.0; P = 0.01) were associated with a EULAR good clinical response, whereas type of biologic drug, disease duration, DAS28 scores, HAQ scores, VAS global health scores, and concomitant use of MTX at baseline showed no statistically significant association with a EULAR good clinical response (all P > 0.05).

In a similar analysis using the ACR20 response as the dependent variable, concomitant use of MTX (OR 1.7 for use versus nonuse, 95% CI 1.1–2.6; P = 0.03), a high VAS score for global health (OR 1.01/cm increase, 95% CI 1.00–1.02; P = 0.01), and a CRP level >10 mg/liter (OR 2.4 for high versus low, 95% CI 1.7–3.9; P < 0.001) were associated with clinical response. The unadjusted effect of a high CRP level was an OR of 2.6 (95% CI 1.7–3.9; P < 0.001).

In a multiple logistic regression analysis with ACR50 as the dependent variable, a CRP level >10 mg/liter was the only variable associated with clinical response (OR 3.0, 95% CI 2.0–4.5; P < 0.001). A similar result was found in the model using ACR70 as the measure of clinical response (for high CRP, OR 3.6, 95% CI 2.2–5.9; P < 0.001). Sex, age, type of biologic drug, disease duration, DAS28 scores, HAQ scores, VAS global health scores, and concomitant MTX use were not associated with ACR50 or ACR70 improvement responses.

When patients were stratified according to their baseline level of CRP (≤10 mg/liter versus >10 mg/liter), the percentages of patients achieving a clinical response within 6 months of treatment were as follows: 51% versus 73% achieving ACR20, 33% versus 60% achieving ACR50, 14% versus 37% achieving ACR70, and 47% versus 65% achieving a EULAR good response. Thus, 1 in 7 patients achieved clinical improvement according to the ACR70 if their CRP level was in the normal range at treatment start, whereas 1 in 3 patients whose CRP level was elevated at treatment start achieved an ACR70 improvement response.

DISCUSSION

TNFα inhibitors are efficacious in the treatment of patients with PsA in randomized clinical trials (22–26). Data from clinical practice are needed to complement the trials and to assess their external validity (27). Published studies on patients with PsA treated with biologic drugs are, however, few (4, 8, 9, 28, 29). The present nationwide study of 764 patients with PsA followed up prospectively for up to 8 years in routine care represents the largest cohort and the longest observation time published to date. Three findings of this study are of importance for clinical practice: approximately half of the patients with PsA who received their first treatment series with a TNFα inhibitor achieved a clinical response (ACR50 or EULAR good response) within 6 months. The number needed to treat (NNT) to achieve an ACR70 response was much higher in patients whose baseline CRP level was in the normal range (NNT = 7) than in patients whose baseline CRP level was elevated (>10 mg/liter) (NNT = 3). CRP was the only baseline variable that was predictive of both longer treatment continuation and good treatment response.

Carmona et al reported a 1-year drug survival rate of 88% among 570 patients with PsA registered in the Spanish national registry, although no efficacy data were reported (8). Similarly, 76% of British patients with PsA followed up for more than 1 year were continuing to receive treatment with their first TNFα inhibitor, according to British registry data (5). A report based on data from the South Swedish register did not explicitly present the drug survival rates, but a EULAR good response rate of 55% was reported among 261 patients with PsA (9), a rate that was very similar to the 54% found in our study. These findings are difficult to compare with the results from randomized controlled trials, as those trials have often focused on ACR20 clinical responses or shorter observation periods, but the 6-month ACR50 response rate of 45% in our study is similar to the rates of 40–45% previously reported in randomized trials (22–26).

An increased CRP level at baseline was the sole factor most uniformly linked to clinical response and treatment continuation. Similar results have been reported by other authors among patients with PsA as well as patients with RA and patients with spondylarthropathies (9, 30–37). The CRP level is probably linked to systemic inflammation in patients with rheumatic disease, and thus it might be used to distinguish those patients who mainly experience chronic irreversible damage from those with active inflammatory disease. In contrast, some authors have found high baseline HAQ or VAS scores to be predictive of drug discontinuation, perhaps because of an association with chronic irreversible disease (31). It would be of great value to select the patients most likely to benefit from biologic treatment at early stages of the decision process; in this aspect, CRP levels seem to be a promising candidate.

We found that MTX use was associated with treatment continuation, a tendency that has been previously described by others (5, 9, 38) but only in analyses that were adjusted for other baseline variables. MTX use was not a uniform predictor of clinical response in our study, since it was only a predictor of the ACR20 response and not the EULAR response or the ACR50 or ACR70 response. These discrepancies may be explained, at least in part, by the concept of bias by indication with regard to concomitant MTX treatment. No data were available on the reasons for addition of MTX to the biologic treatment regimen in some patients and not in others. Absence of concomitant MTX use might be associated with the presence of a comorbidity of importance for drug continuation (9). MTX use was linked to use of infliximab, but it is currently unknown whether MTX has any effect on the formation of immunopathogenic antibodies in PsA during anti-TNFα therapy (39). Randomized controlled trials of anti-TNF medications in PsA have not shown any beneficial effect associated with the use of MTX as a concomitant treatment in patients with PsA (23, 24).

In the present study, men had a longer treatment duration and better EULAR response than did women (5, 28, 36, 38). Shorter treatment duration and poorer treatment response among women treated with TNFα inhibitors have also been described in patients with RA and in those with ankylosing spondylitis (37, 38, 40, 41). Similarly, the tendency toward higher VAS and HAQ scores in women has previously been observed (41, 42). However, any linkage between musculoskeletal performance, sex hormones, or other sex-related factors and anti-TNFα therapy has not been established (41).

The main reason for drug discontinuation was lack of treatment effect, whereas only a few patients stopped treatment due to adverse events. Discontinuation of infliximab use was related to adverse events. A similar tendency has been described previously (5, 43), but, due to the observational study design and lack of randomization of drug therapy, such data must be interpreted with caution.

The DANBIO data in our study can be considered of high quality. According to previous reports, >90% of Danish patients treated with biologic agents are registered in the DANBIO database, probably due to the fact that registration is mandatory irrespective of patient's consent (13). This is supported by the fact that coverage is much lower in databases using voluntary registration and requiring patient consent (8). In the DANBIO registry, similar to that in most clinical trials, the type of arthritis, peripheral or axial, is not registered. Since the disease entities are treated differently (44), this may have influenced the results. Similarly, smoking status and comorbid disease are other factors that might influence treatment outcomes (5). However, these data were only recently introduced in the registry and are still not uniformly available.

This analysis of 764 patients with PsA in a nationwide prospective registry documents that TNFα inhibitors decrease the disease activity in patients with PsA in clinical practice. Overall, half of the patients achieved a good clinical response, but the response rate and duration of drug survival were significantly higher in patients with an elevated CRP level at baseline. Male sex was predictive of longer treatment continuation and a EULAR good response. The effect of concomitant MTX use was weak but seemed to have some beneficial effect on drug survival and achievement of the ACR20 response. Other parameters, including the type of biologic drug, HAQ score, DAS28 score, VAS score, and disease duration, did not significantly affect drug survival or treatment efficacy. Among these Danish patients with PsA, the TNFα-inhibiting treatments were well tolerated, and only a few patients stopped treatment due to adverse effects.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Glintborg had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Glintborg, Østergaard, Tarp, Hetland.

Acquisition of data. Glintborg, Dreyer, Tarp, Rifbjerg-Madsen, Lorenzen, Hetland.

Analysis and interpretation of data. Glintborg, Østergaard, Dreyer, Krogh, Tarp, Hansen, Hetland.

Acknowledgements

We thank all of the rheumatology departments in Denmark for reporting patient data to the DANBIO registry.

Ancillary