Funding sources The cost of the DERMBIO database maintenance is covered by unrestricted grants of equal size from the manufacturers of biologic agents [Abbott, Janssen-Cilag, Merck-Serono (2007–2009), MSD, Pfizer] but the sponsors have no influence on data collection, analysis or interpretation of the data and have no access to the data.
Conflicts of interest R.G. has obtained research grants from Abbott and lecture fees from Abbott, Janssen-Cilag, MSD and Pfizer. K.K. is consultant and/or speaker for Abbott, Basilea, Janssen-Cilag, LeoPharma, MSD, Novo Nordisk and Pfizer. T.N.D. is consultant and/or speaker for Abbott, Janssen-Cilag, MSD, LeoPharma, Novo Nordisk and Pfizer. L.S. is consultant and/or speaker for Abbott, Janssen-Cilag, MSD, LeoPharma and Pfizer.
Background Adherence to treatment is an indicator of treatment success. Long-term data on adherence to biologic treatment in psoriasis are lacking.
Objectives To compare the tumour necrosis factor (TNF)-α inhibitors regarding drug survival rate and safety in patients with psoriasis.
Methods This study is based on data from the Danish nationwide database DERMBIO covering patients with psoriasis treated with a biologic agent. All patients who received anti-TNF-α treatment in academic referral centres were included. Baseline data, adverse events, time on treatment and reason for stopping treatment were recorded. Hazard ratios (HRs) for factors determining drug survival were calculated by logistic regression.
Results In total, 882 treatment series with etanercept (n =311), adalimumab (n =427) or infliximab (n =144) were administered to 747 patients. Significant predictors of drug survival were: sex, the anti-TNF-α agent and the previous response to an anti-TNF-α agent. In the group of anti-TNF-α-naïve patients the longest drug survival was observed for infliximab, followed by adalimumab [HR vs. infliximab 3·70, 95% confidence interval (CI) 1·99–6·89] and etanercept (HR vs. infliximab 3·18, 95% CI 1·72–5·86). The 4-year drug survival is in the range of 40% for etanercept or adalimumab vs. 70% for infliximab. There was no difference in number of adverse events.
Conclusions The overall efficacy of anti-TNF-α drugs diminishes with time, as envisaged by the progressive loss of patient adherence to treatment. The major reasons for stopping treatment were loss of efficacy, followed by adverse events. Infliximab had the best patient retention ability, with 70% of patients still being on the drug after 4 years of treatment.
Introduction of the biologic agents in dermatology challenged the established principles of psoriasis management. In contrast to the traditional systemic agents such as ciclosporin or methotrexate in which the rotational treatment principle has been used due to the risk of the cumulative end-organ toxicity, biologic agents are used for long-term treatment. Currently there is no accepted limit for the duration of the treatment and the clinical objective is long-term psoriasis control using continuous treatment with a single agent. The feasibility of this approach has been shown in relatively short clinical trials, which demonstrated high efficacy and satisfactory tolerability of the biologic agents. There is a particular large experience with the use of the antitumour necrosis factor (TNF)-α agents such as adalimumab, etanercept or infliximab. Sixteen randomized clinical trials documented that 50–90% of patients are likely to obtain a satisfactory reduction in the severity of disease [as measured by a ≥ 75% decrease in Psoriasis Area and Severity Index (PASI) compared with baseline] and that the treatment is well tolerated over a period of 1–1·5 years.1 However, safety and efficacy data for longer treatment durations (> 2 years) are lacking.2
Adherence to treatment is an overall marker of treatment success, as it depends on drug efficacy, side-effects and patients’ satisfaction with the treatment.3–5 Low adherence increases the cost of treatment and lack of adherence is one of the most important factors which increases the cost of therapy.6 Quantification of patient retention in therapy and identification of factors improving drug adherence are therefore important. Adherence of patients with psoriasis to the anti-TNF-α agents decreases with time. A recent review of the data from 19 clinical trials has shown that 5–29% of patients discontinue the anti-TNF-α agent within the 24–30-week treatment period.7 Adherence to treatment (also referred to as ‘drug survival’) seems to be highest for etanercept followed by adalimumab and infliximab. Data for longer treatment durations are scarce but suggest an annual drop-out rate in the range 15–25%.7
As long-term, randomized, controlled trials comparing different anti-TNF-α agents with respect to efficacy and treatment adherence have not been performed, the observational studies are the only source of information to assess the potential differences between the three TNF-α inhibitors regarding their ability to provide a long-term effect for patients with psoriasis. The aim of the present study was therefore to quantify the adherence of patients with psoriasis to treatment with etanercept, adalimumab and infliximab and to identify clinical prognostic factors for long-term drug survival. This study is based on data from the Danish nationwide clinical database DERMBIO covering patients with psoriasis treated with biologic agents.
Patients and methods
The DERMBIO database
The DERMBIO database was started May 2007 due to an initiative of Danish dermatologists. The aim is to monitor the efficacy and side-effects of biologic drugs in patients with psoriasis and to control the quality of the treatment across different centres. The registry is internet based and uses open-source software similar to the Danish rheumatological database, Danbio.8 The cost of database maintenance is covered by unrestricted grants of equal size from the manufacturers of biologic agents [Abbott, Janssen-Cilag, Merck-Serono (2007–2009), MSD, Pfizer] but the sponsors have no influence on data collection and have no access to the data. The DERMBIO database is administered by an independent steering committee consisting of specialists in dermatology representing the five university departments of dermatology and the private dermatology practices in Denmark. All patients with a dermatological disease treated with biologic drugs in Denmark should be registered in the database. Reporting is done at each visit, as a minimum at initiation of treatment, at the 3-month visit and then at every 12 months of follow-up.
Danish guidelines for the treatment of psoriasis with biologic agents
To be eligible for biologic treatment in Denmark the patients must fulfil the Danish national guidelines which are: diagnosis of psoriasis vulgaris; for biologic-naïve patients the disease severity documented as PASI > 10 or Dermatology Life Quality Index (DLQI) > 10 or affected body surface area > 10%; lack of efficacy to methotrexate and narrowband ultraviolet B or in case of intolerance to methotrexate lack of efficacy to another systemic treatment (ciclosporin or acitretin).
Data covering individual treatment series for each anti-TNF-α agent were extracted from the database into an Excel file. Only patients treated in the academic centres were included. The data set contained: (i) patient identification number; (ii) sex; (iii) age; (iv) diagnosis and duration of psoriasis; (v) the identity of the anti-TNF-α agent; (vi) presence of psoriasis arthritis; (vii) presence of comorbidities by the following categories: hypertension, ischaemic heart disease, hypercholesterolaemia, diabetes, nonmelanoma skin cancer, alcohol abuse; (viii) previous anti-TNF-α treatment and reason for discontinuation; (ix) duration of the treatment series; (x) reason for treatment termination; (xi) PASI and DLQI score at baseline (2 weeks before or after start of treatment) and at week 13, week 52, last visit or end of treatment; (xii) side-effects; (xiii) concomitant treatment with methotrexate if the patient received more than one dose of the drug during the treatment series; (xiv) number of previous systemic treatments; and (xv) duration of disease.
Data were analysed using the following software: Microsoft Excel and GraphPad Prism version 5.00 for Mac (GraphPad Software, San Diego, CA, U.S.A.) (basic statistics) and SPSS statistical package (SPSS Inc., Chicago, IL, U.S.A.) for the Cox regression analysis. For the Cox regression analysis the data were stratified by drug, sex, concomitant methotrexate treatment and by previous anti-TNF-α treatment (anti-TNF-α -naïve patients, or previous anti-TNF-α treatment discontinued because of lack of efficacy). The results are presented as Kaplan–Meier survival curves showing the probability for drug survival together with hazard ratio (HR) calculated using the proportional hazards model for drug withdrawal. P <0·05 was considered significant.
The dataset consisted of 882 treatment series with etanercept, adalimumab or infliximab administered to 747 patients. Of these, 311 treatment series were with etanercept, 427 with adalimumab, and 144 with infliximab. The patients were followed until the database lock on 18 February 2010 or until they withdrew from treatment, whichever came first. Patient characteristics are presented in Table 1. There were no significant differences between patients treated with different anti-TNF-α agents according to the parameters listed in Table 1.
Table 1. Characteristics of patients with psoriasis in the study population
Adalimumab (n =347)
Etanercept (n =271)
Infliximab (n =129)
Total (n =747)
PASI, Psoriasis Area and Severity Index; CI, confidence interval; DLQI, Dermatology Life Quality Index. aNo significant difference between adalimumab, etanercept and infliximab groups, P >0·05 (anova).
Males/females, n (% males)a
Age (years), median (range)a
PASI baseline, mean (95% CI)a
Disease duration (years), mean (95% CI)a
DLQI baseline, mean (95% CI)a
Patients with psoriasis arthritis (%)a
Concomitant methotrexate, n (%)a
Withdrawal from treatment, n
Lack of efficacy
Predictors of adherence to the antitumour necrosis factor-α agents
We performed Cox regression analysis to determine the overall drug survival in patients treated with the anti-TNF-α agents. The analysis was done on 715 patients, for whom the recorded time of observation was longer than 0 months. As 206 patients received two or more treatments and had observation time exceeding 0 months, the same patient could be included twice or thrice in the analysis, each time for a different treatment series with another anti-TNF-α agent. The significant positive predictors of drug survival were: male sex, the anti-TNF-α agent and the previous response to the anti-TNF-α agent (Fig. 1). Other tested parameters, including age, duration of psoriasis, PASI or DLQI score at baseline, presence of psoriasis arthritis, concomitant methotrexate, presence of any comorbidity or presence of the metabolic comorbidities (hypertension, diabetes, ischaemic heart disease or hypercholesterolaemia) did not have any influence on drug survival.
Drug survival rate for adalimumab, etanercept and infliximab
As shown in Figure 1, the strongest predictor of drug survival was the previous failure of an anti-TNF-α agent. To compare the three anti-TNF-α agents we have therefore performed separate Cox regression analyses for the group of patients who were naïve to the anti-TNF-α agents (n =564 informative cases) and the patients who experienced lack of efficacy of one or more anti-TNF-α agent(s) (n =151 informative cases) (Fig. 2). In the group of anti-TNF-α-naïve patients the longest drug survival was observed for infliximab, followed by adalimumab [HR vs. infliximab 3·70, 95% confidence interval (CI) 1·99–6·89] and etanercept (HR vs. infliximab 3·18, 95% CI 1·72–5·86). The difference between infliximab survival and etanercept or adalimumab was statistically significant (P <0·0001), but there was no difference between etanercept and adalimumab. The 4-year drug survival is in the range of 40% for etanercept or adalimumab vs. 70% for infliximab. The situation is different in the subgroup of patients who previously failed one or more anti-TNF-α agent(s). Here, the drug survival was very poor compared with the anti-TNF-α -naïve patients (4-year survival < 10%) with only minimal differences between the different drugs. Adalimumab tended to demonstrate a slightly poorer survival rate in comparison with infliximab or etanercept (HR 0·64, 95% CI 0·47–0·95). Moreover, in this group there was a weak, but statistically significant, trend towards improvement of survival in patients who received concomitant therapy with methotrexate (P =0·01, HR 0·655). As data in Figure 2b were generated by analysing treatment series rather than individual patients (i.e. patients who received two different biologics were included twice) we repeated the analysis using only the duration of the second treatment (i.e. treatment after failure of the first anti-TNF-α agent) as an input parameter. The result was comparable with that in Figure 2b and no differences between the drugs were observed (not shown). We have also performed the analogous Cox analysis on the data set including all patients switched from the first anti-TNF-α drug to the second anti-TNF-α agent (n =206 patients with observation period exceeding 1 month), thus including also the patients who discontinued the first treatment for a reason other than lack of efficacy (adverse events or patient-related reasons). Also in this case the survival curves resembled those seen in Figure 2b with no statistical differences between different treatments (not shown).
The DERMBIO database collects the adverse events as grouped under the following headings: allergy, infection, skin rash, systemic lupus eruthematosus-like symptoms and others. Serious adverse events (SAEs) are listed separately. There was a comparable number of adverse events across different anti-TNF-α drugs. Adverse events were recorded in 39·8% of adalimumab treatment series, 39·2% of etanercept treatment series and 38·2% of infliximab treatment series. The most common adverse event registered was infection (no apparent differences in frequency distribution between the three anti-TNF-α drugs; see Table 2). In total, nine SAEs related to anti-TNF-α drugs were recorded at the academic centres, six probably related to the treatment: four in adalimumab-treated patients (one case of ovarian cancer and three cases of serious infection: pneumonia, abscess, urinary tract infection) and two in etanercept-treated patients [one case of neurological symptoms (vertigo) and one case of sepsis].
Table 2. The number of adverse events across different antitumour necrosis factor-α drugs
SLE, systemic lupus erythematosus. aNumber of adverse events in the treatment series: some treatment series had more than one adverse event.
Number of treatment series
Length of all treatment series (patient-years)
Adverse events, n (%)a
Total number of treatment series with adverse events (%)
The major conclusion from our study is that the overall efficacy of anti-TNF-α drugs diminish with time, as envisaged by the progressive loss of patient adherence to treatment. The major reasons for discontinuing the treatment were loss of efficacy (75%), followed by adverse events (12%). The major predictors of poor drug survival were, except for female sex, the type of anti-TNF-α agent and a previous experience of inefficacy with one or more anti-TNF-α agent(s). In our cohort, infliximab had the best patient retention ability, with 70% of patients being still on drug after 4 years of treatment. Adalimumab and etanercept had significantly worse drug survival rates and there was no difference between these two drugs (Fig. 1). These findings are surprising in view of what has been reported for patients treated with anti-TNF-α drugs for rheumatoid arthritis. Studies on patients with rheumatoid arthritis from Denmark,9 Switzerland,10 Germany,11 the U.K.12 and Sweden13 confirmed the inferiority of infliximab to etanercept or adalimumab in the rheumatological setting. The largest and most recent study by Hetland et al.9 reported an infliximab-treated patient retention ability at only 41% after 2 years and documented that drug withdrawals were due mainly to a higher rate of adverse events. This is in marked contrast to the situation found in our psoriasis cohort where the 2-year adherence rate was almost twice as high (79%) and the adverse events were equally distributed among patients treated with the three anti-TNF-α agents. Although the superior patient retention ability needs to be confirmed by other independent registries, there are several possible explanations for our observations. In contrast to rheumatoid arthritis where all three anti-TNF-α agents have comparable efficacy at standard dose,14,15 infliximab is clearly more efficacious for plaque psoriasis than is etanercept or adalimumab.1 A higher overall efficacy in comparison with the alternative drugs may positively influence the motivation of the physician and the patient to stay on the drug. Second, infliximab is the only approved TNF-α blocker which is administered according to patient weight. In contrast to patients with rheumatoid arthritis who weigh less than average, many patients with psoriasis are obese. Although clinical trials with adalimumab and etanercept did not show a significant influence of body weight on treatment responses, there is evidence for worse responses in obese patients16 (reviewed by Clark and Lebwohl17). It is conceivable that the subcutaneous anti-TNF-α agents given in a predetermined dose are less efficacious in those with high body weight. Unfortunately, patient weight is not routinely reported to the DERMBIO registry so this hypothesis remains unconfirmed. Third, the intervals for infliximab dosages can be shortened according to patient response18 but this option is not available in the guidelines for adalimumab and etanercept where constant dosing is recommended for all patients. Another significant predictor of drug survival was female sex. The reason(s) for why women exhibit shorter times of treatment retention than men can only be a subject of speculation, but this has previously been reported for patients with psoriasis arthritis.12 A far more important predictive factor was a previous failure of one or more anti-TNF-α agent(s) (HR 5·3, Fig. 1). Even though one could expect lower efficacy of a drug in patients who were unresponsive to another drug from the same class, this phenomenon has never been conclusively shown for anti-TNF-α blockers and is a subject for discussion. It has been argued that treatment failure with one anti-TNF-α agent does not predict response to another anti-TNF-α agent.16,19–21 In patients with rheumatoid arthritis a clear benefit of switching from one anti-TNF-α agent to another has been demonstrated,22 but other studies indicated a significantly lower response rate for the second or third TNF-α inhibitor.23 A retrospective study of 49 patients, of whom 37 had previously been exposed to another anti-TNF-α agent, showed sustained responses to adalimumab.21 In a very small prospective study on five patients who failed high-dose etanercept (100 mg weekly), four patients showed a ≥ 50% decrease in PASI by week 12 on treatment with double-dose adalimumab (40 mg weekly).19 However, both studies were seriously underpowered and were not designed to test long-term adherence to treatment.
Data of Hetland et al.9 indicate that patients who are concomitantly treated with methotrexate have a significantly better drug survival rate. This treatment is likely to decrease drug catabolism and to prevent formation of drug autoantibodies. Our analysis did not show any effect of concomitant methotrexate in the whole treated population. However, subgroup analysis revealed a modest, but significant, improvement of drug survival in patients who previously failed an anti-TNF-α blocker and received methotrexate in the subsequent treatment series. This analysis should be interpreted with caution, as only 7·8% of all patients in the DERMBIO registry were recorded to receive methotrexate. Secondly, data on the dose and duration of concomitant methotrexate treatment have not been collected, so it is not possible to assess the effect of this drug in a more rigorous way.
Analyses of data collected from registries have inherent deficiencies. Although the data in DERMBIO have been acquired prospectively, the patients were not randomized to different treatments, introducing a selection bias. It is, however, reassuring that the baseline characteristics of the patients do not seem to differ between the different treatment groups. A potential source of error is that the baseline data were collected within a 4-week window (± 2 weeks) in relation to treatment start. Thus the patients who experienced a rapid effect of the biologic could have lower PASI and DLQI values at baseline.
Our findings that patient retention in anti-TNF-α therapy varies between different anti-TNF-a agents and is reduced in patients who have previously failed another TNF-α blocker have practical implications. Long-term psoriasis control is not only a desirable clinical objective, but also a sound strategy from the pharmacoeconomic perspective. Initiating a therapy with a drug that eventually will lose its efficacy increases the overall cost of therapy. It will therefore be worthwhile to explore further the reasons for the superiority of infliximab in patients with psoriasis and further to optimize the treatment with the subcutaneously administered anti-TNF-α agents.
What’s already known about this topic?
• Antitumour necrosis factor (TNF)-α agents are effective for treatment of psoriasis for a short period.
What does this study add?
• Patients with psoriasis treated with anti-TNF-α agents for a longer period show progressive loss of adherence to treatment, mainly due to loss of efficacy.
The DERMBIO study group comprised a representative for the private dermatology practices in Denmark and the departments of dermatology at the following hospitals in Denmark: Bispebjerg, Århus, Roskilde, Odense and Gentofte. Data were extracted from the DERMBIO database by Mikkel Abildtoft, Zitelab Aps.