Atopic dermatitis (AD) is a chronic, inflammatory skin disease characterized by intense pruritus and recurrent dry, red and excoriated patches.1 AD is associated with sleep disturbance, anxiety, depression and work absenteeism,2-4 which places a substantial burden on patients and society.5, 6

Management of AD is aimed at long-term control of the disease, preventing flares and avoiding adverse events (AEs).4 According to guidelines and consensus statements, patients with AD who fail or are intolerant of topical treatments should be treated systemically with immunosuppressants or biologics.4, 7-9 Therapies currently used for the systemic treatment of moderate-to-severe AD include immunosuppressants, such as corticosteroids, ciclosporin A (CSA), methotrexate and azathioprine, which may be associated with toxicities, in addition to one biologic, dupilumab, and a small-molecule inhibitor of Janus kinase, baricitinib.4, 10 Topical corticosteroids (TCS) and topical calcineurin inhibitors are commonly used in combination with these therapies; however, their efficacy is affected by patient adherence. Achieving the right balance of efficacy and safety is a critical factor in patients’ preferences in relation to treatment.11, 12 The range of systemic treatment options that provide sustained disease control and have an acceptable safety profile is still limited.7, 8, 13-15

CSA is considered a first-line treatment for patients with severe AD in some countries; however, it is approved for use only in Europe and Japan and has limitations to its use.4, 8 CSA use is associated with side-effects and requires close monitoring for signs of hypertension and nephrotoxicity; therefore, continuous treatment beyond 2 years is not recommended and some suggest a treatment duration of only 3–6 months.4, 8 With limited treatment options, patients with severe AD who have inadequate response with CSA or are ineligible for CSA are an important patient population to investigate.

Tralokinumab is a fully human immunoglobulin G4 monoclonal antibody that specifically targets the interleukin-13 cytokine, a key driver of the signs and symptoms of AD.16, 17 Tralokinumab, used concomitantly with TCS or as monotherapy, was effective and well tolerated with a favourable benefit–risk profile in phase II and III trials in patients with moderate-to-severe AD.18-20

The aim of this trial was to evaluate the efficacy and safety of tralokinumab plus TCS as needed in a European population of adult patients with severe AD whose disease was not adequately controlled with CSA or who had contraindications to oral CSA.

Patients and methods Trial design

ECZTRA 7 (ClinicalTrials.gov NCT03761537; EU Clinical Trials Register 2018-000747-76) was a randomized, double-blind, placebo-controlled, parallel-group, 26-week, phase III clinical trial (Figure 1). The trial was conducted across 68 sites (Appendix S1; see Supporting Information).

Trial design. AD, atopic dermatitis; q2w, every 2 weeks; TCI, topical calcineurin inhibitor; TCS, topical corticosteroids.

Patients

The trial enrolled adults (aged ≥ 18 years) with a diagnosis of AD (defined by Hanifin and Rajka criteria)21 for ≥ 1 year, who had an inadequate response to treatment with topical or documented systemic medications in the past year, and whose disease was not adequately controlled with CSA or who had contraindications to oral CSA use (Appendix S2; see Supporting Information). Patients were required to have AD involvement of ≥ 10% of body surface area, an Eczema Area and Severity Index (EASI) score of ≥ 20 and an Investigator’s Global Assessment (IGA) score of ≥ 3 at screening and at baseline, and worst daily pruritus numeric rating scale (NRS) average score of ≥ 4 during the week before baseline.

Ethics

This trial was sponsored by LEO Pharma A/S and was conducted in accordance with ethical principles originally described in the Declaration of Helsinki and Good Clinical Practice guidelines, and was approved by the local institutional review boards and independent ethics committees of each institution. All patients provided signed, informed consent before any trial procedure.

Study procedures, randomization and treatments

Patients entered a screening period, including a washout, of 2–6 weeks, depending on the type of prior medications (Appendix S2; see Supporting Information). The use of TCS (any potency) was permitted, but not mandated, during screening up to randomization. Patients were randomized 1 : 1 to receive 26 weeks of subcutaneous tralokinumab 300 mg every 2 weeks plus TCS as needed, following a 600-mg loading dose on day 0, or placebo every 2 weeks plus TCS as needed. Details on randomization and blinding are provided in Appendix S2 (see Supporting Information). After completion of the visit at week 26, eligible patients were invited to enter a long-term extension trial (ECZTEND; NCT03587805). Patients who did not transfer to ECZTEND completed a 14-week off-treatment follow-up for safety, pharmacokinetics and antidrug antibodies (Figure 1).

All patients were instructed to apply a thin layer of supplied TCS once daily as needed (mometasone furoate 0·1% cream, provided free of charge in kit sizes of 180–255 g every 2 weeks) to areas with active lesions. The safety and appropriateness of continued repeated courses of TCS were monitored and use was discontinued gradually when control was achieved. Patients were instructed to return used and unused tubes at each visit to allow measurement of the amount of TCS that had been used. Patients were instructed to apply an emollient twice daily (or more as needed) for at least 14 days before randomization and throughout the trial. Rescue treatment (topical or systemic medications) was permitted to control intolerable AD symptoms at the investigator’s discretion for the duration of the trial. Patients discontinued trial treatment if they received systemic corticosteroids or nonsteroidal systemic immunosuppressive drugs, but they could resume trial treatment after treatment with these medications, if deemed appropriate by the investigator.

Endpoints

The primary endpoint was EASI 75 at week 16. Secondary endpoints were reduction of worst daily pruritus NRS (weekly average) of ≥ 4 from baseline, change from baseline in SCORing Atopic Dermatitis (SCORAD) score and Dermatology Life Quality Index (DLQI) scores (both at weeks 16 and 26) and EASI 75 at week 26. Other endpoints included EASI 90 and change in EASI, Patient-Oriented Eczema Measure (POEM), eczema-related sleep, amount of TCS used and number of days without topical treatment use, all from baseline to weeks 16 and 26.

Statistical analyses

A sample size of 250 patients was estimated to provide 99% power for detecting a treatment difference between tralokinumab and placebo with a two-sided 5% significance level, assuming a response rate for EASI 75 at week 16 of 40% and 15%, respectively.

To control the overall type 1 error rate at a 5% significance level, testing for statistical significance for primary and secondary endpoints was prespecified and carried out in sequential order following an endpoint hierarchy (Appendix S3; see Supporting Information).

A range of statistical analyses was prespecified within the estimand framework as per current guidelines,22 incorporating three types of intercurrent events that could influence the treatment effect, i.e. ‘initiation of rescue medication’, ‘permanent discontinuation of treatment’ and ‘patient onset of the COVID-19 pandemic’.

For binary endpoints, a ‘COVID-19 modified composite’ estimand was specified as the primary approach, assessing the treatment difference in response rates achieved without rescue medication and/or discontinuation of the investigational medicinal product.

Observed data not influenced by intercurrent events were used ‘as is’. For patients with missing data owing to COVID-19 restrictions or with ‘patient onset of the COVID-19 pandemic’ as a first intercurrent event prior to week 16 or week 26, any missing data or observed data after the event were not used but assumed missing at random and imputed using multiple imputation. Patients with missing data not resulting from COVID-19 restrictions or who were affected by either of the other intercurrent events were considered nonresponders. The difference in response rates between treatment groups was analysed for each imputed dataset using the Cochran–Mantel–Haenszel test stratified by prior CSA use (yes/no) and baseline IGA score (3/4). Estimates and standard errors were combined using Rubin’s rule to form a unique point estimate and standard error.

For continuous endpoints, a ‘hypothetical’ estimand was applied, assessing treatment difference in change from baseline to week 16 or week 26, as if all patients adhered to treatment with no intercurrent events. Data collected after the first intercurrent event were considered missing and endpoints were analysed using a linear mixed model for repeated measurements. Details on model, sensitivity and other estimand analyses for binary and continuous endpoints are provided in Appendix S3 (see Supporting Information).

Safety analyses are reported up to week 26 for the safety analysis set. The COVID-19 restrictions influenced trial events and data collection, and changes to the protocol were introduced as mitigation (Appendix S3; see Supporting Information). Data interpretation and conclusions have not been affected.

Post hoc analyses

EASI 75 response in a subgroup of patients with CSA treatment failure (defined as inadequate effect after > 12 weeks of exposure or AE reporting related to CSA) and EASI ≤ 7 from week 16 to week 26 were analysed as for the prespecified binary endpoints, but with modified stratification factors. The analysis of 5–7 days per week without topical treatment use considered the intercurrent events ‘initiation of rescue medication’ and ‘permanent discontinuation of treatment’. Patients with missing data or who were affected by intercurrent events were imputed as nonresponders. The difference in response rates was analysed using the Cochran–Mantel–Haenszel test stratified by prior CSA use and baseline IGA score. The analysis of cumulative TCS use applied similar methods as specified for the continuous endpoints (Appendix S3; see Supporting Information).

Results Patient disposition and baseline characteristics

Between December 2018 and September 2020, 318 patients were screened and 277 patients were randomized; of these, two patients did not receive either treatment, leading to analysis being conducted in 275 patients (tralokinumab, N = 138; placebo, N = 137) (Figure 1, Figure S1; see Supporting Information). Baseline demographics and disease characteristics were similar across treatment groups, with a median duration of AD of 26 years and a median body surface area involvement of 52% (Table 1, Table S1; see Supporting Information). All patients had severe AD at baseline (EASI score ≥ 20); median EASI score was 28·8 and median SCORAD score was 69·1 (Table 1). All patients had received previous AD therapies (Table 1). Atopy history and medical history at baseline are provided in Tables S2 and S3 (see Supporting Information).

Table 1. Patient demographics and baseline characteristics All randomized (N = 277) Placebo q2w + TCS (N = 137) Tralokinumab q2w + TCS (N = 140) Median age, years 34·0 (26·0–45·0) 34·0 (26·0–45·0) 33·0 (25·5–47·0) Male, n (%) 165 (59·6) 83 (60·6) 82 (58·6) Median weight, kg 74·0 (63·0–86·0) 75·0 (63·0–88·0) 74·0 (63·0–86·0) Median BMI, kg m−2 24·96 (21·80–28·09) 25·43 (21·62–28·95) 24·84 (22·10–27·72) Race, n (%) White 272 (98·2) 135 (98·5) 137 (97·9) Black or African American 1 (0·4) 1 (0·7) 0 Asian 1 (0·4) 1 (0·7) 0 Other 3 (1·1) 0 3 (2·1) Median duration of AD, years 26·0 (18·0–34·5) 25·0 (17·0–34·0) 26·0 (18·0–35·0) Median BSA involvement, % 52·0 (35·0–70·0) 52·0 (35·0–70·0) 52·0 (36·5–70·0) Median EASI score 28·8 (22·4–39·6) 29·1 (22·8–40·2) 28·6 (22·4–38·0) Median SCORAD total score 69·1 (61·4–78·9) 68·9 (61·2–81·0) 69·2 (61·5–76·5) Median DLQI score 16·0 (11·0–21·0) 16·0 (11·0–21·0) 16·0 (11·0–21·0) Median POEM total score 22·0 (17·0–26·0) 22·0 (17·0–26·0) 22·0 (18·0–26·0) Median weekly average of worst daily pruritus NRS 7·4 (6·6–8·3) 7·5 (6·6–8·4) 7·4 (6·4–8·3) Previous CSA use, n (%) 207 (74·7) 102 (74·5) 105 (75·0) Reason for discontinuation of CSA treatment, n (%) Inadequate efficacy 95 (34·3) 43 (31·4) 52 (37·1) Side-effects 86 (31·0) 45 (32·8) 41 (29·3) Treatment duration > 1 year 12 (4·3) 7 (5·1) 5 (3·6) Other 14 (5·1) 7 (5·1) 7 (5·0) No previous CSA use, n (%) 70 (25·3) 35 (25·5) 35 (25·0) Reason for no previous CSA use, n (%) Risk of side-effects 9 (3·2) 4 (2·9) 5 (3·6) Contraindications 52 (18·8) 24 (17·5) 28 (20·0) Other 9 (3·2) 7 (5·1) 2 (1·4) Previous treatments for AD, n (%) Topical corticosteroids 276 (99·6) 136 (99·3) 140 (100) Systemic steroids 189 (68·2) 91 (66·4) 98 (70·0) Mycophenolate 8 (2·9) 5 (3·6) 3 (2·1) Methotrexate 49 (17·7) 26 (19·0) 23 (16·4) Azathioprine 36 (13·0) 18 (13·1) 18 (12·9) Monoclonal antibody 21 (7·6) 12 (8·8) 9 (6·4) Dupilumab 14 (5·1) 8 (5·8) 6 (4·3) Other immunosuppressant 28 (10·1) 12 (8·8) 16 (11·4) Phototherapy 163 (58·8) 84 (61·3) 79 (56·4) AD, atopic dermatitis; BMI, body mass index; BSA, body surface area; CSA, ciclosporin A; DLQI, Dermatology Life Quality Index; EASI, Eczema Area and Severity Index; NRS, numeric rating scale; POEM, Patient-Oriented Eczema Measure; q2w, every 2 weeks; SCORAD, SCORing Atopic Dermatitis; TCS, topical corticosteroids. Data are provided as median (interquartile range) unless otherwise stated.

Overall, 207 patients (74·7%) were previously exposed to CSA for AD prior to trial entry; the main reasons for CSA treatment discontinuation were inadequate efficacy (34·3%) and side-effects (31·0%), and the main reason patients did not receive prior CSA treatment was contraindication (18·8%) (Table 1).

Rescue medication was used by fewer patients receiving tralokinumab compared with placebo (5·7% vs. 13·9%) (Table S4 and Figure S2; see Supporting Information). Higher-potency TCS was the most commonly used rescue medication. Systemic rescue treatments, most commonly systemic corticosteroids, were used by 2·1% of patients receiving tralokinumab and 8·0% of patients receiving placebo (Table S4; see Supporting Information).

Most patients in the tralokinumab plus TCS group and placebo plus TCS group (89·3% and 87·6%, respectively) completed the trial up to week 26. One patient in the tralokinumab plus TCS group (0·7%) withdrew from treatment owing to lack of efficacy, and no patients withdrew in the placebo plus TCS group. Two patients (one from each treatment group) withdrew owing to site closure caused by COVID-19 restrictions. Four patients [three (2·2%) from the placebo plus TCS group and one (0·7%) from the tralokinumab plus TCS group] withdrew from treatment before week 26 owing to AEs. Patients with missing data or who were affected by intercurrent events, including those related to COVID-19, are summarized in Table S5 (see Supporting Information).

Efficacy Primary endpoint

At week 16, the proportion of patients achieving EASI 75 was significantly higher in the tralokinumab plus TCS group vs. the placebo plus TCS group [64·2% vs. 50·5%; difference 14·1%, 95% confidence interval (CI) 2·5–25·7; P = 0·018] (Table 2, Figure 2a).

Table 2. Efficacy outcomes Responders

Placebo q2w + TCS

(N = 137)

Tralokinumab q2w + TCS

(N = 140)

Treatment difference (95% CI)a P-values Primary efficacy outcome, n/N (%) EASI 75 at week 16b,c,d 69·2/137 (50·5) 88·6/138 (64·2) 14·1 (2·5–25·7) 0·018 Secondary efficacy outcomes EASI 75 at week 26, n/N (%)b,c,d 75·7/137 (55·3) 95·0/138 (68·8) 14·1 (2·9–25·3) 0·014 EASI 90 at week 16, n/N (%)b,c,d 40·2/137 (29·3) 56·7/138 (41·1) 12·3 (1·1–23·6) 0·032 EASI 90 at week 26, n/N (%)b,c,d 49·8/137 (36·4) 67·1/138 (48·6) 12·9 (1·4–24·4) 0·027 EASI 50 at week 16, n/N (%)b,c,d 95·3/137 (69·5) 110·4/138 (80·0) 10·6 (0·3–20·8) 0·043 EASI 50 at week 26, n/N (%)b,c,d 91·9/137 (67·1) 111·2/138 (80·5) 13·7 (3·5–23·9) 0·008 Adjusted mean change from baseline in weekly average of worst daily pruritus NRS at week 16 (± SE)e,f n = 136; –3·1 (0·2) n = 137; –4·0 (0·2) –0·9 (–1·4 to –0·4) < 0·001 Adjusted mean change from baseline in weekly average of worst daily pruritus NRS at week 26 (± SE)e,f n = 136; –3·4 (0·2) n = 137; –4·3 (0·2) –0·9 (–1·4 to –0·3) 0·002 Weekly average of worst daily pruritus NRS reduction of ≥ 4 at week 16b,c,d n = 48·0/135 (35·6) n = 61·0/134 (45·5) 9·7 (–2·0 to 21·4) 0·106 Weekly average of worst daily pruritus NRS reduction of ≥ 4 at week 26b,c,d n = 53·7/135 (39·7) n = 63·3/134 (47·2) 7·3 (–4·6 to 19·2) 0·228 Adjusted mean change from baseline in SCORAD score at week 16 (± SE)e,f n = 137; –34·1 (1·6) n = 138; –42·7 (1·6) –8·6 (–13·0 to –4·2) < 0·001 Adjusted mean change from baseline in SCORAD score at week 26 (± SE)e,f n = 137; –37·3 (1·6) n = 138; –46·3 (1·5) –8·9 (–13·2 to –4·6) < 0·001 Adjusted mean change from baseline in EASI at week 16 (± SE)e,f n = 137; –22·4 (0·8) n = 138; –26·4 (0·8) –3·9 (–6·3 to –1·6) < 0·001 Adjusted mean change from baseline in EASI at week 26 (± SE)e,f n = 137; –23·7 (0·8) n = 138; –27·2 (0·8) –3·5 (–5·7 to –1·3) 0·002 Adjusted mean change from baseline in DLQI at week 16 (± SE)e,f n = 134; –9·6 (0·4) n = 137; –11·2 (0·4) –1·5 (–2·6 to –0·4) 0·009 Adjusted mean change from baseline in DLQI at week 26 (± SE)e,f n = 134; –9·9 (0·4) n = 137; –11·5 (0·4) –1·6 (–2·7 to –0·5) 0·005 Adjusted mean change from baseline in POEM at week 16 (± SE)e,f n = 134; –8·3 (0·6) n = 135; –11·7 (0·6) –3·4 (–5·0 to –1·8) < 0·001 Adjusted mean change from baseline in POEM at week 26 (± SE)e,f n = 134; –9·1 (0·6) n = 135; –12·6 (0·6) –3·6 (–5·3 to –1·9) < 0·001 Adjusted mean change from baseline in eczema-related sleep NRS at week 16 (± SE)e,f n = 136; –3·4 (0·2) n = 137; –4·1 (0·2) –0·8 (–1·3 to –0·2) 0·005 Adjusted mean change from baseline in eczema-related sleep NRS at week 26 (± SE)e,f n = 136; –3·7 (0·2) n = 137; –4·3 (0·2) –0·6 (–1·1 to –0·0) 0·037 CI, confidence interval; CSA, ciclosporin A; DLQI, Dermatology Life Quality Index; EASI, Eczema Area and Severity Index; IGA, Investigator’s Global Assessment; N, number of patients; n, number of patients with observation; NRS, numeric rating scale; POEM, Patient-Oriented Eczema Measure; q2w, every 2 weeks; SCORAD, SCORing Atopic Dermatitis; SE, standard error; TCS, topical corticosteroids. aMantel–Haenszel risk difference stratified by prior CSA use and baseline IGA. bIntercurrent events used: rescue treatment and permanent discontinuation of the investigational medicinal product (neither resulting from the COVID-19 pandemic) and patient onset of the COVID-19 pandemic. cPatients who received rescue treatment or permanently discontinued the investigational medicinal product as their first intercurrent event before the analysed visit were considered nonresponders. dData collected after patient onset of the COVID-19 pandemic were considered missing. Data missing as a result of the COVID-19 pandemic were imputed using multiple imputation. Missing data that were not attributable to the COVID-19 pandemic were imputed as nonresponse. eData collected after permanent discontinuation of the investigational medicinal product, initiation of rescue treatment or patient onset of the COVID-19 pandemic were not included. fOutcome measures were produced using least square means with observed margins as weighting. Repeated-measurements model: change from baseline = treatment × week + baseline value × week + prior CSA use + country (Germany: yes/no) + baseline IGA. If no postbaseline scheduled assessments had taken place before any intercurrent event, the last observation was carried forward (baseline included).

Efficacy outcomes by visit. (a) Achievement of Eczema Area and Severity Index (EASI) 75 by visit from baseline to week 26. (b) Change from baseline in SCORing Atopic Dermatitis (SCORAD) score by visit. (c) Reduction in worst daily pruritus numeric rating scale (NRS) (weekly average) of ≥ 4 by visit from baseline to week 26. (d) Achievement of EASI 90 by visit from baseline to week 26. (e) Change from baseline in Dermatology Life Quality Index (DLQI) score by visit. Change from baseline data were produced using least square means with observed margins as weighting from a repeated-measurements model [change = treatment × week + baseline value × week + prior ciclosporin A use + country (Germany: yes/no) + baseline Investigator’s Global Assessment]. Data collected after permanent discontinuation of investigational medicinal product, initiation of rescue treatment or patient onset of the COVID-19 pandemic were not included. If no postbaseline scheduled assessments had taken place before any intercurrent event, the last observation was carried forward (baseline included). For binary analyses, intercurrent events used included rescue treatment and permanent discontinuation of investigational medicinal product (neither resulting from the COVID-19 pandemic) and patient onset of the COVID-19 pandemic. Patients who received rescue treatment or permanently discontinued investigational medicinal product as their first intercurrent event before the analysed visit were considered nonresponders. Data collected after patient onset of the COVID-19 pandemic were considered missing. Data missing as a result of the COVID-19 pandemic were imputed using multiple imputation. Patients with missing data that was not a result of the COVID-19 pandemic were imputed as nonresponders. *Nominal 5% statistical significance. q2w, every 2 weeks; SE, standard error; TCS, topical corticosteroids.

Secondary and other endpoints

For the first secondary endpoint in the prespecified testing hierarchy, a greater proportion of patients achieved a reduction in worst daily pruritus NRS (weekly average) of ≥ 4 from baseline to week 16 with tralokinumab plus TCS vs. placebo plus TCS (45·5% vs. 35·6%); however, this did not reach statistical significance (difference 9·7, 95% CI −2·0 to 21·4; P = 0·106) (Table 2) and subsequent secondary endpoints were interpreted nominally.

The adjusted mean change in SCORAD score from baseline to week 16 was greater for tralokinumab plus TCS vs. placebo plus TCS (−42·7 vs. −34·1; difference −8·6, 95% CI −13·0 to −4·2; P < 0·001) and continued to increase up to week 26 (Table 2, Figure 2b). A greater proportion of patients also achieved a reduction in worst daily pruritus NRS (weekly average) of ≥ 4 at week 26 with tralokinumab plus TCS vs. placebo plus TCS (Table 2, Figure 2c). The proportion of patients achieving EASI 75 with tralokinumab plus TCS further increased beyond week 16 until week 26 (68·8% vs. 55·3%; difference 14·1, 95% CI 2·9–25·3; P = 0·014) (Figure 2a). The adjusted mean change in EASI from baseline was greater for tralokinumab plus TCS vs. placebo plus TCS at week 16 (P < 0·001) and week 26 (P = 0·002). From week 16 to week 26, more patients treated with tralokinumab plus TCS achieved the clinically meaningful threshold of an absolute EASI ≤ 723, 24 vs. placebo plus TCS (48·0% vs. 32·6%; difference 16·2, 95% CI 4·9–27·5; P = 0·005) (Table 3). Likewise, the percentage of patients achieving EASI 90 at week 16 was greater with tralokinumab plus TCS vs. placebo plus TCS (41·1% vs. 29·3%; difference 12·3, 95% CI 1·1–23·6; P = 0·032), further increasing up to week 26 without plateauing (Table 2, Figure 2d).

Table 3. Post hoc efficacy outcomes Responders Placebo q2w + TCS (N = 137) Tralokinumab q2w + TCS (N = 140) Treatment difference (95% CI)a P-values EASI ≤ 7 from week 16 to week 26, n/N (%)a 44·7/137 (32·6) 66·3/138 (48·0) 16·2 (4·9–27·5) 0·005 5–7 days per week without topical treatment use (eDiary) at week 16, n/N (%)b,c 28/133 (21·1) 57/130 (43·8) 22·9 (11·9–33·8) < 0·001d