The association between etanercept serum concentration and psoriasis disease severity is poorly investigated, and currently etanercept serum concentration monitoring that is aiming to optimize the psoriasis treatment lacks evidence. In this prospective study, we investigated the relation between etanercept exposure and disease severity via measuring etanercept concentrations at five consecutive time points in 56 psoriasis patients. Disease severity assessments included the Psoriasis Area and Severity Index (PASI), body surface area (BSA) and Physician Global Assessment (PGA), and etanercept and anti-etanercept antibody concentrations were determined every 3 months for a period of 1 year. The present study demonstrated that the association between etanercept concentration and psoriasis severity is age-dependent: when patients were stratified into three groups, patients in the youngest age group (–50 years) showed a lower PASI at a higher etanercept concentration (β = –0.26), whereas patients in the oldest age group (+59 years) showed the opposite trend (β =0.22). Similar age effects were observed in the relation of etanercept concentration with BSA (P=0.02) and PGA (P=0.02). The influence of age and length of time in therapy on the etanercept concentration–disease severity relation was unaffected by body mass index (BMI) or any other possible confounder. Incidence of anti-etanercept antibodies was low (2%). The age-dependent relation between etanercept serum concentrations is both unexpected and intriguing and needs further investigation.

## Introduction

Psoriasis is a recurrent chronic immune-mediated inflammatory disorder of the skin, affecting 2–3% of the population worldwide. The disease has a strong genetic component but environmental factors such as infections can play an important role in the presentation as well [1]. Psoriasis affects the skin with red plaques covered with whitish scales. In addition to an impairment of quality of life, patients with psoriasis run a higher risk of developing psoriatic arthritis and cardiovascular disease [1,2].

Given that psoriasis most often strikes between the ages of 15 and 35, patients require lifelong treatment. Consequently, costs associated with psoriasis present a considerable economic burden [3]. In case of moderate-to-severe disease, biological agents targetting distinct inflammatory pathways are usually indicated. Etanercept, a fully human tumour necrosis factor α (TNF) soluble receptor fusion protein that antagonizes the effects of endogenous TNF, is a well-established biological treatment for psoriasis [4].

Although the standard approved dosing regimens of biologic agents have been established in large randomized controlled trials, clinical experience in real-life practice settings suggests that off-label dose adjustments (e.g. dose escalations, interval changes or interruptions) are frequently applied and therefore clinically relevant. The reasons for these off-label uses are various and include improvement of drug efficacy, addressing drug tolerance and preparing patients for surgery avoiding significant infectious risks [5]. In order to guide clinical decision making in a more objective way, therapeutic drug monitoring (TDM) using trough concentration measurements is emerging as an important tool to sustain decision making in clinical practice.

The main goal of TDM is to keep trough concentrations within a well-defined therapeutic window, so that the drug can exert its full effect and immunogenicity may be prevented, but without overexposure. Indeed, a number of recent reviews have highlighted the association between low infliximab and adalimumab serum trough concentrations, the development of antidrug antibodies and worse disease outcomes [6,7]. For etanercept, however, satisfying results were obtained using both continuous and interrupted treatment regimens [8,9], putting the role of serum etanercept and anti-etanercept antibody measurements into question. Moreover, new reimbursement criteria approved by Belgian authorities in 2011 allow treatment of moderate-to-severe psoriasis with flexible (continuous or intermittent) etanercept dosing, making this biologic currently the only one to offer label-approved flexible treatment tailored for patient need [10].

In this report, we have sought to explore the relation between etanercept dosing and disease severity in a cohort of 56 psoriasis patients via measurement of etanercept exposure. In doing so, we have extended the etanercept dose–effect relation to age-dependent differences and identified other important factors that need to be considered.

## Experimental

### Study design

#### Patients, treatment regimen and disease severity scores

This prospective observational cohort study consisted of 56 patients with plaque psoriasis treated with etanercept at the Department of Dermatology of the University Hospital of Leuven. All subjects enrolled met the new reimbursement criteria for etanercept and could be treated concomitantly with acitretin or methotrexate from the start of the study onwards. Patients received etanercept 50 mg subcutaneously every week (continuous group) or had a treatment stop of at least 2 weeks because of a good response (intermittent group). In patients with inadequate response to etanercept, concomitant acitretin could be started or the dose of etanercept could be increased to 50 mg subcutaneously twice weekly. Decisions were made by the treating dermatologist and did not deviate from daily practice. Disease severity was assessed during the patients’ routine clinic consultations using the Psoriasis Area and Severity Index (PASI), body surface area (BSA) and Physician Global Assessment (PGA) every 3 months for a period of 1 year. The PASI, BSA and PGA can vary from 0 to 72, 0 to 100% and 0 to 5 respectively, with higher scores indicating more severe conditions. The treating dermatologist was blinded for the serum etanercept and anti-etanercept antibody status of the patients. Demographic data were recorded from medical histories and patient medical records. Laboratory tests included C-reactive protein (CRP) determination. The present study was approved by the Ethical Committee of the University Hospitals Leuven (B322201422128/S56818) and all patients gave written informed consents.

#### Blood samples

Blood samples were taken during the patients’ routine clinic consultations just before the next administration of drug (in order to obtain an etanercept trough concentration). Blood was collected in ‘BD Vacutainer SST II Advance tube’ (BD Biosciences), containing a gel separator and clot activator. After an incubation time of 30 min, serum was prepared by centrifugation (10 min, 1960×g, room temperature) and stored at –20°C until analysis.

### Laboratory analysis

#### Measurement of serum etanercept concentrations

Serum trough concentrations of etanercept were determined by the in-house developed MA-ETN63C8/MA-ETN61C1-HRP ELISA [11]. The assay cutoff was set at 0.2 µg/ml taking into account a serum dilution factor of 1:1000. Assessment of interference of endogenous soluble TNF receptor in the detection of etanercept using the MA-ETN63C8/MA-ETN61C1-HRP ELISA revealed a 100-fold lower reactivity of endogenous soluble TNF receptor compared with etanercept [11].

#### Measurements of serum anti-etanercept antibody concentrations

Anti-etanercept antibodies were detected by means of an in-house developed bridging ELISA using MA-ETN64A5 as calibrator [11]. Cutoff level for a positive signal was set at 3.2 ng/ml equivalents MA-ETN64A5 taking into account a serum dilution factor of 1:20. However, as this type of assay does not detect anti-etanercept antibodies bound to etanercept and therefore probably underestimates antidrug antibody formation, a newly developed drug-tolerant assay (based on the ‘affinity capture elution’ method by Bourdage et al. [12]) was used as well. In short, possible immune complexes of etanercept and anti-etanercept antibody in samples were dissociated by treatment with glycine buffer (pH 2.5) and then added to Tris (pH 9.5) buffered etanercept-coated plates. After washing, bound anti-etanercept antibody was eluted by addition of glycine buffer (pH 2.5) and acid eluate was transferred to fresh Tris (pH 9.5) buffered plates. Plates were incubated and blocked with 1% BSA in PBS (pH 7.4). Plate-bound anti-etanercept antibody was subsequently detected by the addition of biotin-labelled etanercept as primary conjugate and HRP-labelled streptavidin as secondary conjugate. Colour development was performed using o-phenylenediamine and H2O2 in citrate buffer (pH 5.0). The measured values were reported as positive when the concentration was above 25 ng/ml equivalents MA-ETN64A5 (serum dilution factor of 1:18).

### Statistical analysis

Quantitative data were summarized by mean and standard deviation (mean (S.D.)) or median and interquartile range (median (IQR)), as appropriate. Frequencies were compared using the Fisher’s exact test. Linear mixed effects modelling was performed to determine whether trough etanercept concentration and disease severity of the psoriasis patients were mutually correlated. The dependent variable was either disease severity or trough etanercept concentration measured at five consecutive visits (3 months apart). We started with a model with fixed effects ‘comorbidity’, ‘CRP’, ‘comorbidity × CRP’, ‘starter’, ‘body mass index (BMI)’, ‘starter × BMI’, ‘trough etanercept’, ‘starter × trough etanercept’, ‘age (at study entry)’, ‘trough etanercept × age’, ‘time’, ‘trough etanercept × time’ and ‘treatment schedule’ for PASI as dependent variable and a model with fixed effects ‘starter’, ‘BMI’, ‘PASI’, ‘PASI × starter’, ‘age (at study entry)’, ‘PASI × age’, ‘treatment schedule’, ‘treatment schedule × age’, ‘time’, ‘PASI × time’ for trough etanercept concentration as dependent variable. Selection was based on factors associated with the dependent variable identified one by one (P-value<0.05). Model building was performed based on the general guidelines for model construction, outlined by Verbeke et al. [13]. Model comparison was based on Akaike’s Information Criterion (AIC) or minus twice the Restricted Log Likelihood ratio test, wherever appropriate. Our final model consisted of a random intercept and random linear trend (time) and accounted for within-individual dependency via a first-order autoregressive covariance structure. Selection of the remaining fixed effected was Restricted Maximum Likelihood (REML) based and valid if dropout was missing at random. Effect sizes were presented as parameter estimates (β). To identify and determine the effects of variables associated with an intermittent etanercept regimen, binary logistic regression analysis was performed. Effects were considered to be statistically significant when P<0.05. All statistics were carried out using the statistical programs Graphpad Prism 5.0 (GraphPad Software, San Diego, CA) and IBM SPSS Statistics 23 (IBM SPSS, Costa Mesa, CA).

## Results

Demographics and disease characteristics of the cohort of psoriasis patients at the time of enrolment are shown in Table 1. This cohort included 56 patients (of whom nine had just started on etanercept therapy), 39 male and 17 female, with a mean (S.D.) age of 53 [11] years and a median (IQR) disease duration of 25 [13–36] years. Nearly half of the patients (45%) received concomitant treatment with acitretin and approximately one-third (34%) had a co-diagnosis of psoriatic arthritis. At study entry, median (IQR) PASI, BSA and PGA were 3.3 (2.7–4.2), 2.9 (1.4–4.5) and 2 (1–2) respectively, indicating an overall good response rate to etanercept therapy.

Table 1
Summary characteristics of the psoriasis patients at study entry
n=56
Age (years) 53 (11)
Gender: male, n (%) 39 (70%)
BMI (kg/m226.9 [24.4–31.2]
Disease duration (years) 25 [13–36]
Concomitant acitretin, n (%) 25 (45%)
Methotrexate, n (%) 3 (5%)
Previous anti-TNF use, n (%) 12 (21%)
Starters, n (%) 9 (16%)
Psoriatic arthritis, n (%) 19 (34%)
Cardiovascular comorbidities, n (%) 23 (41%)
Days on etanercept therapy 1263 [487–1636]
CRP (mg/l) 1.1 [0.6–2.8]
PASI 3.3 [2.7–4.2]
BSA 2.9 [1.4–4.5]
PGA 2 [1–2]
n=56
Age (years) 53 (11)
Gender: male, n (%) 39 (70%)
BMI (kg/m226.9 [24.4–31.2]
Disease duration (years) 25 [13–36]
Concomitant acitretin, n (%) 25 (45%)
Methotrexate, n (%) 3 (5%)
Previous anti-TNF use, n (%) 12 (21%)
Starters, n (%) 9 (16%)
Psoriatic arthritis, n (%) 19 (34%)
Cardiovascular comorbidities, n (%) 23 (41%)
Days on etanercept therapy 1263 [487–1636]
CRP (mg/l) 1.1 [0.6–2.8]
PASI 3.3 [2.7–4.2]
BSA 2.9 [1.4–4.5]
PGA 2 [1–2]

Values are expressed as median [interquartile range] or as mean (S.D.), unless stated otherwise.

### Etanercept concentration – disease severity – effect modification

Although there was no association between etanercept concentration and disease severity as such (etanercept concentration–PASI: β = –0.04, P=0.59; etanercept concentration–BSA: β =0.12, P=0.39; etanercept concentration–PGA: β = –0.01, P=0.81), striking associations were observed within different age groups, which were statistically highly significant. For example, a significant age interaction was observed in the relation of etanercept concentration with PASI score (interaction, P=0.01): when patients were stratified into three age groups equal in size (22–50 years; 51–58 years and 59–75 years), patients in the youngest age group showed a lower PASI at a higher etanercept concentration (β = –0.26), patients in the oldest age group showed the opposite trend (β =0.22) (Figure 1). Similar interaction effects were observed in the relation of etanercept concentration with BSA (interaction, P=0.02) and in the relation of etanercept concentration with PGA (interaction, P=0.02). These age interactions render the relation between etanercept concentration and disease severity (and thereby therapeutic response to etanercept) age-dependent.

#### Effect (from linear mixed effects modelling) on PASI score, of unit difference in etanercept concentration (1 µg/ml) modelled for all three age groups.

Figure 1
Effect (from linear mixed effects modelling) on PASI score, of unit difference in etanercept concentration (1 µg/ml) modelled for all three age groups.

$estimated PASI from a random intercept and random linear trend (time) model, accounted for within- individual dependency via a first-order autoregressive covariance structure. *p<0.05 Figure 1 Effect (from linear mixed effects modelling) on PASI score, of unit difference in etanercept concentration (1 µg/ml) modelled for all three age groups.$estimated PASI from a random intercept and random linear trend (time) model, accounted for within- individual dependency via a first-order autoregressive covariance structure. *p<0.05

Interestingly, patients in the oldest age group (+59 years) were more significantly co-treated with acitretin (65% compared with 34%, P=0.04) and had a longer disease duration (32 years compared with 24 years, P=0.02) compared with patients in the younger groups (<59 years). Other differences (old compared with younger) included the percentage of patients that were obese (BMI ≥30 kg/m2) (12% compared with 39%, P=0.04) and the percentage of patients in the intermittent cohort (50% compared with 11%, P=0.01). Patients in the youngest age group (–50 years), by contrast, had more often just started on etanercept therapy (37% compared with 6%, P=0.01) compared with the older groups (>50 years).

### Etanercept concentration – disease severity – treatment regimen

Of the 56 psoriasis patients, 39 (70%) were treated continuously with etanercept and 12 (21%) patients received an intermittent regimen. It was not possible to determine the nature of etanercept treatment in 5 (9%) patients as they prematurely ended the study.

In evaluable patients, mean (S.D.) PASI was significantly lower in the intermittent cohort compared with the continuous cohort (2.6 (0.8) compared with 3.5 (1.5)) (β = –0.91, P=0.01). The same was true for the other markers of disease severity: mean (S.D.) BSA was 1.9 (1.2) compared with 3.4 (2.8) (β = –1.44, P=0.02) and mean (S.D.) PGA was 1.5 (0.6) compared with 1.9 (0.6) (β = –0.45, P=0.00) in the intermittent compared with continuous cohort respectively.

With respect to the etanercept concentration, no difference was found between patients who were either continuously (mean (S.D.) =3.7 (1.6) µg/ml) or intermittently (mean (S.D.) =3.3 (1.3) µg/ml) treated with etanercept (P=0.36). However, also in this case, an age-dependent interaction effect (interaction, P=0.01) was clearly observed (Figure 2). For patients in the intermittent cohort (who were generally doing better), the etanercept concentration was lower at older age (β = –0.09); for patients in the continuous cohort, on the other hand, a higher etanercept concentration was observed at the older age (β =0.04).

#### Effect (from linear mixed effects modelling) on etanercept concentration, of unit difference in age (1 year) modelled for the two treatment groups.

Figure 2
Effect (from linear mixed effects modelling) on etanercept concentration, of unit difference in age (1 year) modelled for the two treatment groups.

$estimated etanercept concentration from a random intercept and random linear trend model, accounted for within-individual dependency via a first-order autoregressive covariance structure *p<0.05 Figure 2 Effect (from linear mixed effects modelling) on etanercept concentration, of unit difference in age (1 year) modelled for the two treatment groups.$estimated etanercept concentration from a random intercept and random linear trend model, accounted for within-individual dependency via a first-order autoregressive covariance structure *p<0.05

Binary logistic regression analysis revealed that for every additional year in age, the likelihood of being treated intermittently had increased by 11% (P=0.02). Other factors like gender, absence of psoriatic arthritis, use of acitretin and a low BMI were not significantly associated with an intermittent therapy regimen.

### Etanercept concentration – disease severity – adjustment for other factors

To adjust the age-dependent etanercept concentration–response relation for other factors, influence of gender, BMI, psoriatic arthritis, acitretin, CRP, cardiovascular comorbidity (hypertension, hypercholesterolaemia, hyperlipidaemia, diabetes or metabolic syndrome), disease duration and previous anti-TNF use on both etanercept concentration and main disease severity parameter (i.e. PASI) was investigated. Four important factors could be distinguished:

#### BMI

BMI of the psoriasis patients significantly affected both etanercept concentration and PASI score. With regard to the etanercept concentration, an increase in one unit in BMI (1 kg/m2) was associated with a decrease in etanercept concentration of approximately 0.1 µg/ml (β = –0.07, P=0.02). Mean (S.D.) etanercept concentration in patients who were obese compared with patients who were not was 2.8 (1.1) µg/ml compared with 3.9 (1.6) µg/ml (β = –0.96, P=0.02) respectively. The opposite trend was observed for the PASI score: an increase in one unit in BMI (1 kg/m2) was associated with an increase in PASI of approximately 0.1 (β =0.08, P=0.00). Mean (S.D.) PASI in patients who were obese compared with patients who were not was 3.8 (1.7) compared with 3.0 (1.3) (β =1.05, P=0.00) respectively.

#### CRP-cardiovascular comorbidity

No association was found between CRP and PASI score or CRP and etanercept concentration respectively. However, taking into account the presence of cardiovascular comorbidity, a significant interaction effect was observed in the relation of CRP with PASI (interaction, P=0.01): in patients with cardiovascular comorbidity, CRP and PASI were much more positively correlated than in patients without cardiovascular comorbidity (β =0.27 compared with β =0.01 respectively) (Figure 3). Mean (S.D.) CRP was not elevated (>5mg/l) and did not differ between patients with and without cardiovascular comorbidity (1.5 (1.6) mg/l compared with 2.2 (2.7) mg/l, P=0.08). There was also no difference in PASI score (3.3 (1.5) compared with 3.3 (1.4), P=0.82) between both patient groups, nor was there any difference in etanercept concentration (3.5 (1.6) µg/ml compared with 3.7 (1.5) µg/ml, P=0.69) between patients with and without cardiovascular comorbidity.

##### Effect (from linear mixed effects modelling) on PASI score, of unit difference in CRP (1 mg/l) modelled for patients with and without cardiovascular comorbidity.
Figure 3
Effect (from linear mixed effects modelling) on PASI score, of unit difference in CRP (1 mg/l) modelled for patients with and without cardiovascular comorbidity.

$estimated PASI from a random intercept and random linear trend (time) model, accounted for within- individual dependency via a first-order autoregressive covariance structure. *p<0.05 Figure 3 Effect (from linear mixed effects modelling) on PASI score, of unit difference in CRP (1 mg/l) modelled for patients with and without cardiovascular comorbidity.$estimated PASI from a random intercept and random linear trend (time) model, accounted for within- individual dependency via a first-order autoregressive covariance structure. *p<0.05

#### Starters on etanercept therapy

Of the 56 patients enrolled, nine (16%) had just started on etanercept therapy (inclusion start date – treatment start date ≤3 months). Mean (S.D.) PASI was significantly higher in starters than in the rest of the psoriasis patients (3.9 (1.9) compared with 3.1 (1.3)) (β =0.91, P=0.02). As a result, a smaller increase in etanercept concentration and/or a smaller decrease in BMI much more positively affected the PASI score (Table 2). There was no difference in mean (S.D.) etanercept concentration (3.5 (1.3) µg/ml compared with 3.6 (1.6) µg/mL) (P=0.81) between both the patient groups.

Table 2
Factors independently associated with PASI score or etanercept trough concentration, determined by linear mixed effects modelling
PASI score Etanercept trough concentration
Variable β-coefficient (95% CI) P-value Variable β-coefficient (95% CI) P-value
Intercept 0.25 (–3.75–4.25) 0.90 Intercept 11.72 (6.75–16.69) 0.00
No-starter 3.73 (–0.18–7.64) 0.06 Continuous schedule –5.33 (–10.59–(–0.07)) 0.04
BMI1 0.26 (0.16–0.37) 0.00 Age at study entry –0.11 (–0.20–(–0.02)) 0.01
No-starter*BMI1 –0.22 (–0.33-(-0.10)) 0.00 Continuous*age 0.10 (0.01–0.19) 0.03
Trough etanercept1 –1.16 (–1.78–(–0.54)) 0.00 PASI1 –1.14 (–1.70–(–0.57)) 0.00
No-starter*trough1 0.66 (0.32–1.00) 0.00 No-starter*PASI1 0.36 (0.12–0.60) 0.00
Age at study entry –0.05 (–0.08–(–0.01) 0.01 No-starter –1.11 (–2.48–0.26) 0.11
Age*trough1  0.04 Age*PASI1  0.00
e.g. 22–50 years –0.37 (–0.59–(–0.15))   e.g. 22–50 years –0.29 (–0.37–(–0.21))
e.g. 59–75 years 0.13 (0.02–0.23)   e.g. 59–75 years 0.34 (0.27–0.40)
BMI1 –0.07 (–0.12–(–0.01)) 0.03
Time1 0.17 (0.08–0.25) 0.00
PASI score Etanercept trough concentration
Variable β-coefficient (95% CI) P-value Variable β-coefficient (95% CI) P-value
Intercept 0.25 (–3.75–4.25) 0.90 Intercept 11.72 (6.75–16.69) 0.00
No-starter 3.73 (–0.18–7.64) 0.06 Continuous schedule –5.33 (–10.59–(–0.07)) 0.04
BMI1 0.26 (0.16–0.37) 0.00 Age at study entry –0.11 (–0.20–(–0.02)) 0.01
No-starter*BMI1 –0.22 (–0.33-(-0.10)) 0.00 Continuous*age 0.10 (0.01–0.19) 0.03
Trough etanercept1 –1.16 (–1.78–(–0.54)) 0.00 PASI1 –1.14 (–1.70–(–0.57)) 0.00
No-starter*trough1 0.66 (0.32–1.00) 0.00 No-starter*PASI1 0.36 (0.12–0.60) 0.00
Age at study entry –0.05 (–0.08–(–0.01) 0.01 No-starter –1.11 (–2.48–0.26) 0.11
Age*trough1  0.04 Age*PASI1  0.00
e.g. 22–50 years –0.37 (–0.59–(–0.15))   e.g. 22–50 years –0.29 (–0.37–(–0.21))
e.g. 59–75 years 0.13 (0.02–0.23)   e.g. 59–75 years 0.34 (0.27–0.40)
BMI1 –0.07 (–0.12–(–0.01)) 0.03
Time1 0.17 (0.08–0.25) 0.00

CI, confidence interval; NS, not significant.

1included as time-varying covariates.

Table 2 shows that the age-dependent and ‘starter’ effect of etanercept concentration on PASI score (and vice versa) remained significant after adjustment of above-mentioned factors using linear mixed effects modelling. In addition, we tested whether different responses and trough levels in different age groups were not simply related to different disease durations and different lengths of etanercept treatment. Therefore, we first checked for multicollinearity among age, disease and treatment duration, which seemed not to be of relevance (variance inflation factor =1). Next, we tested whether disease and/or treatment duration modified the etanercept concentration–psoriasis severity relation in our cohort of patients. Here, we did indeed find a significant interaction (P=0.01) with regard to treatment duration: for patients with a shorter treatment duration, a higher etanercept concentration was associated with a lower PASI score (especially in the first 3 years of etanercept therapy), whereas for patients with a longer treatment duration (5 years or longer on etanercept therapy), the opposite trend was once again observed (results not shown). However, after adjustment for the other factors in the final model, only ‘the first year’ on etanercept treatment remained significantly associated with the PASI score, which is in-line with the already defined ‘starter’ effect of the etanercept concentration–psoriasis severity relation.

### Anti-etanercept concentration – etanercept concentration – disease severity

Patient samples were also analysed for presence of anti-etanercept antibodies using the in-house developed bridging ELISA. In this drug-sensitive assay setup, no antibodies towards etanercept were found in any of the serum samples analysed. The latter also implies that no antibodies could be detected in samples of patients who were intermittently treated with etanercept or in the follow-up samples of a particular patient who had to switch from therapy because of loss of response or in any other patient sample with undetectable serum etanercept trough concentrations. It was therefore decided to reanalyse serum samples of a subgroup of patients with generally a higher risk of having antidrug antibodies (i.e. samples of patients with a trough etanercept concentration ≤2 µg/ml or samples of patients reporting injection site reactions) using the newly developed drug-tolerant assay. In the second format, two samples (from two patients) out of 98 samples (from 40 patients) analysed turned out to be positive. The first positive sample (anti-etanercept concentration of 70 ng/ml equivalents MA-ETN64A5) was derived from a 63-year-old male patient at inclusion in the study (2 months after he had started on etanercept therapy). Trough etanercept concentration at that time was 1.1 µg/ml, PASI 7.3 and BSA 5.5. The patient also reported injection-site reactions. However, after acitretin was added to his treatment schedule, trough etanercept concentration increased to approximately 3 µg/ml, PASI and BSA dropped below a score of 5, and the patient did not report any injection-site reactions anymore. Moreover, anti-etanercept antibodies became undetectable in the drug-tolerant immunoassay as well. The second positive sample (anti-etanercept concentration of 65 ng/ml equivalents MA-ETN64A5) was derived from a 70-year-old female patient who had a serum etanercept trough concentration of 2 µg/ml, a PASI of 5.5 and a BSA of 5.3. Unfortunately, after the first sample, this patient was lost for follow-up.

## Discussion

Our data demonstrate that the association between etanercept concentration and psoriasis severity is age-dependent. For patients in the youngest age group (–50 years), a higher etanercept trough concentration was associated with a better disease outcome, for patients in the oldest age group (+59 years) the opposite was observed. Moreover, the etanercept concentration decreased as age increased in the intermittent dosing cohort, whereas it was higher among older patients in the continuous dosing cohort. The data do not challenge the existing literature, but extend the analysis of etanercept concentration–clinical response relations to age-dependent differences.

How can these differences be explained? First of all, patients in the youngest age group had more often just initiated their etanercept therapy, which was an important determinant of a higher disease activity score. As a consequence, smaller differences in BMI and etanercept concentration more profoundly affected the disease state of such a starting patient. This observation is in-line with the findings reported in a study by Sterry et al. [14], where initial treatment of psoriasis with etanercept 50 mg twice weekly allowed for more rapid clearance of skin lesions than with 50 mg once weekly. However, also in this youngest age group, the age-dependent etanercept concentration–response relation remained significant, even after adjustment of above-mentioned factors (results not shown). Therefore, younger patients, just starting etanercept therapy with a high BMI and a ‘low’ etanercept trough concentration, might benefit from a 50 mg twice weekly etanercept dosing regimen.

Secondly, patients in the oldest age group had a longer disease duration and more frequently belonged to the intermittent cohort. With respect to the latter, an older patient with a lower etanercept trough concentration achieved the best responses. Although this appears to be a paradoxical finding, it is in-line with at least two separate studies for infliximab showing that a low infliximab trough concentration during long-lasting remission is one of the variables associated with a lower risk of relapse after drug withdrawal in patients with inflammatory bowel disease [15,16]. In addition, also for patients in the continuous cohort, the shorter the disease duration but the older the patient, the more etanercept trough concentration and disease outcome were negatively correlated (results not shown).

Why have these age-dependent etanercept concentration–response relations not been reported before? Studies on the value of TDM in psoriasis patients treated with etanercept are limited, the number of included patients small, and – even more important – they mainly focus on determination of clear cutoffs to identify the therapeutic window of etanercept for an entire patient population. In this regard, only one study so far was able to show a positive correlation between etanercept concentration and the percentage decrease in PASI score [17]. However, several studies in other patient groups also investigated the benefit of monitoring serum concentrations of etanercept. Kneepkens et al. [18], for example, demonstrated a significant negative association between etanercept concentration and disease activity score at 24 weeks of therapy in a cohort of 162 consecutive adult patients with ankylosing spondylitis (mean age of 42 years). Patients with low etanercept trough concentrations (<1.8 µg/ml) had a statistically significant higher BMI at baseline compared with patients with high etanercept trough concentrations (≥4.6 µg/ml). Both observations were in-line with the results obtained here.

Although the interaction between CRP and PASI in the presence of cardiovascular comorbidity was not withheld in the final model, it did remain significant and even outweighed the effect of BMI on PASI in the youngest age group of our cohort (results not shown). The reasons why younger psoriasis patients with cardiovascular comorbidity seem to be more prone to the effects of a small CRP (and by extension TNF) elevation are not clear. One study proposed higher soluble TNF receptor levels in the elderly population to oppose the (detrimental) effects of TNF (on development of psoriatic lesions) [19]; however, whether this is the case or other changes in the immune system that occur over time interfere with the anti-TNF mechanism of action, needs further elucidation.

The present study further seems to support the hypothesis that etanercept is not or only marginally immunogenic [20,21]. No antibodies were detected using a drug-sensitive bridging ELISA format, whereas the clinical relevance of low concentration antidrug antibody, typically detected in a drug-tolerant assay only, has recently called into question [22]. Moreover, when compared with adalimumab, a fully human anti-TNF monoclonal antibody demonstrating antidrug antibodies in 20.7% of psoriasis patients whose serum adalimumab concentrations are <2 µg/ml (HUMIRA® Product Monograph), the incidence of antibodies to etanercept in the present study is quite low (approximately 2%). Nevertheless, to the best of our knowledge, the present paper is the first to report anti-etanercept antibodies in psoriasis patients. Initial treatment optimization in one patient confirmed previous findings that the development of antidrug antibodies might be transient and can be overcome [23].

Notably, the obtained results have been derived from relatively small sample sizes, however by acknowledging the longitudinal aspect of the study design allowed to investigate patterns of change along time. Nevertheless, although the associations we report seem statistically robust, they may not be generalizable to other biologics, and they cannot be used to imply direction of causality. The presented study is monocentric and involves a well-defined population of psoriasis patients treated in a similar way, which is crucial to the resolution of dose–effect associations. Nevertheless, most patients were already treated with etanercept for several years and require minimal optimization. Unfortunately, it was not feasible to investigate the success of etanercept dose intensification. Of the five patients in whom dose of etanercept was temporarily increased to 50 mg subcutaneously twice weekly, one 64-year-old male patient had to terminate therapy because of psoriasis exacerbation. For the other four patients, it was difficult to determine whether elevated etanercept trough concentrations after dose escalation for loss of response were associated with improved clinical outcomes, due to the small number of patients involved and the twice weekly dosing regimen (the latter making it impossible to determine an accurate etanercept trough concentration).

In conclusion, the present study provides a new perspective on the role of age as a factor in treatment strategy decision making of etanercept-treated patients with psoriasis. For patients with age below 50 years, an increase in etanercept dose should be considered in case of insufficient control of the psoriasis activity, especially at the start of therapy. For patients above 50, higher doses are not recommended unless etanercept treatment has just been initiated (≤1 year). We, therefore, propose an algorithm (Figure 4) in which age and length on etanercept treatment are combined to identify psoriasis patients eligible for etanercept dose intensification. Another important finding of the present study is that older patients (>60 years) with a long treatment duration and a good response to therapy, might be eligible for an intermittent treatment scheme without any risk to develop anti-etanercept antibodies. Finally, the data suggest that for obese patients, weight loss may positively influence the efficacy of etanercept, which is in agreement with Clark and Lebwohl [24], who showed that fixed-dose biological agents have compromised efficacy in heavier individuals with psoriasis.

### Proposal for an algorithm for psoriasis patients treated with etanercept.

Figure 4
Proposal for an algorithm for psoriasis patients treated with etanercept.

Etanercept-treated patients with psoriasis are stratified according to their age and length of etanercept treatment. For patients with an age below 50 years and/or with short treatment duration (≤1 year), an increase in the etanercept dose should be considered in case of insufficient control of the psoriasis activity.

Figure 4
Proposal for an algorithm for psoriasis patients treated with etanercept.

Etanercept-treated patients with psoriasis are stratified according to their age and length of etanercept treatment. For patients with an age below 50 years and/or with short treatment duration (≤1 year), an increase in the etanercept dose should be considered in case of insufficient control of the psoriasis activity.

## Clinical perspectives

• Measuring serum drug and antidrug antibody concentrations has been proposed as a personalized approach to tailor TNF antagonist therapy. The utility of this concentration-based tailoring in psoriasis patients is, however, unclear.

• In this prospective observational cohort study that included more than 250 serum samples derived from 56 psoriasis patients, the association between etanercept concentration and psoriasis severity was age-dependent, a significant result unaffected by possible confounders.

• The strong effect of age on the etanercept concentration–psoriasis severity association could have a substantial impact on etanercept concentration-based treating algorithms and change the way these patients are managed.

## Funding

This work was supported by the Fund for Scientific Research Flanders [grant number TBM-T003716N]; an Investigator Initiated Research Grant received from Pfizer [grant number WI189787]; and the Agency for the Promotion and Innovation through Science and Technology in Flanders [grant number SB-141478 (to I.D.)].

## Competing interests

S.S. has been a paid speaker or consultant for Abbott, Amgen, Celgene, Janssen, LEO Pharma, Lilly, MSD, Novartis and Pfizer. A.G. has served as a speaker for MSD, Janssen Biologicals, Abbvie, Pfizer and Takeda, as consultant for UCB and Takeda, and has received license of (anti-) infliximab, (anti-) adalimumab, and vedolizumab ELISA to apDia and infliximab, adalimumab lateral flow to R-Biopharm AG. The authors declare that there are no competing interests associated with the manuscript.

## Author contribution

I.D. performed the research, interpreted the data, implemented statistical analysis and drafted the manuscript. K.V.S. helped with the statistical analysis and contributed to manuscript review. S.S. included the patients, provided the serum samples, helped with the interpretation of data and reviewed the manuscript. A.G. designed the research, co-ordinated the experiments and reviewed the manuscript. All authors read and approved the final manuscript.

## Abbreviations

• BMI

body mass index

•
• BSA

body surface area

•
• CRP

C-reactive protein

•
• HRP

•
• IQR

Interquartile Range

•
• PASI

Psoriasis Area and Severity Index

•
• PGA

Physician Global Assessment

•
• TDM

therapeutic drug monitoring

•
• TNF

tumour necrosis factor α

## References

References
1
Griffiths
C.E.M.
and
Barker
J.-N.W.N.
(
2007
)
Pathogenesis and clinical features of psoriasis
.
Lancet
370
,
263
271
2
Langley
R.-G.B.
,
Krueger
G.G.
and
Griffiths
C.E.M.
(
2005
)
Psoriasis: epidemiology, clinical features, and quality of life
.
Ann. Rheum. Dis.
64
,
ii18
ii23
3
Feldman
S.R.
,
Zhao
Y.
,
Navaratnam
P.
,
Friedman
H.S.
,
Lu
J.
and
Tran
M.H.
(
2015
)
Patterns of medication utilization and costs associated with the use of etanercept, adalimumab, and ustekinumab in the management of moderate-to-severe psoriasis
.
J. Manag. Care Spec. Pharm.
21
,
201
209
4
Gottlieb
A.B.
(
2005
)
Psoriasis: emerging therapeutic strategies
.
Nat. Rev. Drug Discov.
4
,
19
34
5
Brezinski
E.A.
and
Armstrong
A.W.
(
2012
)
Off-label biologic regimens in psoriasis: a systematic review of efficacy and safety of dose escalation, reduction, and interrupted biologic therapy
.
PLoS ONE
7
,
e33486
6
T.R.
,
Svenson
M.
,
Eijsbouts
A.M.
,
van den Hoogen
F.H.
,
Enevold
C.
,
van Riel
P.L.
et al.
(
2009
)
Formation of antibodies against infliximab and adalimumab strongly correlates with functional drug levels and clinical responses in rheumatoid arthritis
.
Ann. Rheum. Dis.
68
,
1739
1745
7
Moss
A.
,
Brinks
V.
and
Carpenter
J.F.
(
2013
)
Review article: immunogenicity of anti‐TNF biologics in IBD – the role of patient, product and prescriber factors
.
Aliment. Pharmacol. Ther.
38
,
1188
1197
8
Dauden
E.
,
Griffiths
C.
,
Ortonne
J.P.
,
Kragballe
K.
,
Molta
C.
,
Robertson
D.
et al.
(
2009
)
Improvements in patient‐reported outcomes in moderate‐to‐severe psoriasis patients receiving continuous or paused etanercept treatment over 54 weeks: the CRYSTEL study
.
J. Eur. Acad. Dermatol. Venereol.
23
,
1374
1382
9
Strohal
R.
,
Chimenti
S.
,
Vena
G.A.
and
Girolomoni
G.
(
2013
)
Etanercept provides an effective, safe and flexible short-and long-term treatment regimen for moderate-to-severe psoriasis: a systematic review of current evidence
.
J. Dermatolog. Treat.
24
,
199
208
10
Segaert
S.
,
Ghislain
P.-D.
and
Boone
C.
(
2016
)
An observational study of the real-life management of psoriasis patients treated with etanercept according to the new reimbursement criteria (in Belgium)
.
J. Dermatolog. Treat.
27
,
103
109
11
Detrez
I.
,
Brouwers
E.
,
Peeters
M.
,
Geukens
N.
,
de Vlam
K.
and
Gils
A.
(
2016
)
Characterization and application of a unique panel of monoclonal antibodies generated against etanercept
.
J. Immunol.
196
,
2879
2884
12
Bourdage
J.S.
,
Cook
C.A.
,
Farrington
D.L.
,
Chain
J.S.
and
R.J.
(
2007
)
An Affinity Capture Elution (ACE) assay for detection of anti-drug antibody to monoclonal antibody therapeutics in the presence of high levels of drug
.
J. Immunol. Methods
327
,
10
17
13
Verbeke
G.
and
Molenberghs
G.
(
1997
)
Linear Mixed Models in Practice: A SAS-Oriented Approach
,
Springer-Verlag
,
New York
14
Sterry
W.
,
Ortonne
J.-P.
,
Kirkham
B.
,
Brocq
O.
,
Robertson
D.
,
Pedersen
R.D.
et al.
(
2010
)
Comparison of two etanercept regimens for treatment of psoriasis and psoriatic arthritis: PRESTA randomised double blind multicentre trial
.
BMJ
340
,
c147
15
Ben‐Horin
S.
,
Chowers
Y.
,
Ungar
B.
,
Kopylov
U.
,
Loebstein
R.
,
Weiss
B.
et al.
(
2015
)
Undetectable anti‐TNF drug levels in patients with long‐term remission predict successful drug withdrawal
.
Aliment. Pharmacol. Ther.
42
,
356
364
16
Papamichael
K.
,
Casteele
N.V.
,
Gils
A.
,
Tops
S.
,
Hauenstein
S.
,
Singh
S.
et al.
(
2015
)
Long-term outcome of patients with Crohn’s disease who discontinued infliximab therapy upon clinical remission
.
Clin. Gastroenterol. Hepatol.
13
,
1103
1110
17
Elberdín
L.
,
Outeda
M.
,
P.
,
S.
,
Fernández-Torres
R.M.
,
Iglesias
R.
et al.
(
2016
)
Positive correlation between etanercept concentration and the decrease in Psoriasis Area and Severity Index scale value
.
Int. J. Clin. Pharm.
38
,
1142
1148
18
Kneepkens
E.
,
Krieckaert
C.
,
van der Kleij
D.
,
Nurmohamed
M.
,
van der Horst-Bruinsma
I.
,
Rispens
T.
et al.
(
2015
)
Lower etanercept levels are associated with high disease activity in ankylosing spondylitis patients at 24 weeks of follow-up
.
Ann. Rheum. Dis.
74
,
1825
1829
19
Hasegawa
Y.
,
M.
,
Ozaki
N.
,
Inagaki
T.
and
Suzumura
A.
(
2000
)
Increased soluble tumor necrosis factor receptor levels in the serum of elderly people
.
Gerontology
46
,
185
188
20
de Vries
M.K.
,
van der Horst-Bruinsma
I.E.
,
Nurmohamed
M.T.
,
Aarden
L.A.
,
Stapel
S.O.
,
Peters
M.J.
et al.
(
2009
)
Immunogenicity does not influence treatment with etanercept in patients with ankylosing spondylitis
.
Ann. Rheum. Dis.
68
,
531
535
21
Jullien
D.
,
Prinz
J.C.
and
Nestle
F.O.
(
2015
)
Immunogenicity of biotherapy used in psoriasis: the science behind the scenes
.
J. Invest. Dermatol.
135
,
31
38
22
Detrez
I.
,
Dreesen
E.
,
Van Stappen
T.
,
de Vries
A.
,
Brouwers
E.
,
Van Assche
G.
et al.
(
2016
)
Variability in golimumab exposure: a ‘real-life’ observational study in active ulcerative colitis
.
J. Crohns Colitis
10
,
575
581
23
Casteele
N.V.
,
Gils
A.
,
Singh
S.
,
Ohrmund
L.
,
Hauenstein
S.
,
Rutgeerts
P.
et al.
(
2013
)
Antibody response to infliximab and its impact on pharmacokinetics can be transient
.
Am. J. Gastroenterol.
108
,
962
971
24
Clark
L.
and
Lebwohl
M.
(
2008
)
The effect of weight on the efficacy of biologic therapy in patients with psoriasis
.
J. Am. Acad. Dermatol.
58
,
443
446