Omalizumab for severe atopic dermatitis in 4- to 19-year-olds: the ADAPT RCT

Hypothesis

Our hypothesis is that anti-IgE will reduce the levels of IgE in children with severe eczema and alleviate their symptoms.

Primary objectives

To determine whether or not the intervention is associated with an improvement in eczema severity, compared with placebo.

Secondary objectives

• To evaluate whether or not the intervention is associated with a change in eczema severity, eczema quality of life (QoL) and co-existing allergic disease.

• To address the impact of the intervention on the use of topical drugs in severe eczema, including the use of potent topical steroids and calcineurin inhibitors.

• To compare the drug burden resulting from treatment failure requiring rescue medication with oral steroids, and the need for alternative systemic therapy (AST) (systemic immunosuppression).

• To determine the change in reactivity to allergens.

• To assess the impact of disease burden by assessing the rate of eczema exacerbations and infective episodes of eczema.

Pathophysiology

An interaction of genetics, environment and immunology are implicated in the pathophysiology of eczema. 5

There is 75–85% concordance of eczema in monozygotic twins compared with 30% in dizygotic twins. It is inherited polygenetically, with multiple candidate gene loci proposed. In particular, loss-of-function mutations of filaggrin are associated with early-onset eczema with persistent sensitisation. Filaggrin is involved in the formation of the epidermal barrier through binding to and aggregation of the keratin cytoskeleton, underlying the importance of the skin barrier in eczema and development of allergic sensitisation. Environmental factors trigger exposure to an allergen, leading to a complex progression through sensitisation and ultimately to allergic disease, modulated by environmental factors. Colonisation with Staphylococcus aureus and other organisms can be a result of a reduced innate response. This can, in turn, be a trigger for eczema, possibly related to exotoxin and superantigen effects triggering immune stimulation. Immunological factors, such as T helper 2 (T_h2) cells have been shown to have an important role to play, and are discussed in Emerging systemic therapies. Other irritants and psychoneuroimmunological factors, such as stress, may also play a role in susceptible individuals.

Psychosocial impact

Epidemiology and natural history

Eczema frequently has an onset between 3 and 6 months of age, with 60% of patients developing symptoms within the first year of life. 13 Ten to thirty per cent of patients have residual disease by adulthood, and a minority of patients have adult-onset eczema.

Although only a small proportion of children with eczema are classed as having severe disease,2 this is the subgroup that bears the greatest burden of disease.

Topical treatments

Phototherapy and systemic therapy

Although most patients will respond to topical therapy and supportive measures, there is a small subgroup of patients with severe disease in whom these measures will be unable to control their disease. These children are candidates for phototherapy or systemic therapy.

Emerging systemic therapies

Increasingly, there is a shift towards more targeted therapy, to permit the management of eczema without the attendant side effects of immunosuppression and the long-term consequences of existing treatments. The studies to date have primarily concentrated on new therapies in adult patients.

It has been known for some time that the T_h2 pathway is involved in eczema, and is associated with the upregulation of cytokines interleukin (IL)-4, IL-13 and IL-31. 23

Anti-immunoglobulin E in eczema38,39

There is clearly still a need for a safe and effective treatment for children with severe eczema, who have limited treatment options. There remains a gap in research and in emerging therapies focusing on children with severe eczema. This is an oversight, especially as we know that children suffer greatly with effects that affect the immediate family. In children in particular, the disease can be lifelong, as we have seen that severe eczema in childhood has a tendency to persist. Thus, it is important to identify therapies that can be employed on a long-term basis. In addition, eczema that commences early and persists is strongly associated with asthma and allergic rhinoconjunctivitis. Thus, prevention or treatment of early-onset eczema may reduce the longer-term comorbidity of these other diseases. 40 Omalizumab is the first and only commercially-available anti-IgE antibody. It is a humanised monoclonal antibody that inhibits binding of IgE to the high affinity IgE receptor (FcεRI), thereby limiting mast cell degranulation and the release of inflammatory mediators. 41 The reduced serum-free IgE levels downregulate the high-affinity IgE receptor (FcεRI) surface expression on effector cells, promoting this effect further. It has been approved by the National Institute for Health and Care Excellence (NICE) for the treatment of asthma in patients aged ≥ 6 years and for chronic urticaria in patients aged ≥ 12 years. It is licensed for use from the age of 6 years, as safety data suggest that omalizumab is well tolerated in children in this age range,42 although it has also been used in children from 4 years of age. 43

There is in vitro and murine evidence for the role of IgE in the immunopathogenesis of eczema. Many patients have elevated serum IgE levels and atopy. Furthermore, eczema lesions have been found to bear sizeable numbers of IgE-bearing mast cells, basophils and dendritic cells. They bind the high-affinity receptors (FcεRI), the main IgE-binding structure in eczematous skin. 44 Evidence suggests that the FcεRI-bound allergen-specific IgE presents allergen more effectively to primed T cells,45 leading to T-cell activation and cutaneous inflammation. IgE-mediated histamine release from cutaneous mast cells may also aggravate eczema through the itch–scratch cycle. 46 IgE may play a more important role in children as eczema is thought to become less allergen-driven and more ‘autoreactive’ in adults. This study particularly targets a paediatric atopic population, in which this mechanism may be more relevant. 47,48

Among its many roles, the T_h2 immune pathway drives IgE synthesis. The T_h2 cytokines IL-4, IL-13 and IL-5 lead to IgE class-switching and induce peripheral eosinophils and mast cells. IL-4 and IL-13 are the primary cytokines involved in IgE synthesis and we have seen how dupilumab, which has both anti-IL-4 and IL-13 properties, has been used to manage eczema. Therefore, directly targeting IgE may be a relevant role. Asthma studies have demonstrated that omalizumab can reduce the T_h2 cytokines IL-4, IL-5 and IL-13. 49–51 T_h2 cytokines impair filaggrin and antimicrobial peptide expression,52,53 which play a role in the multifactorial pathogenesis of eczema; this may, therefore, potentially be reversed with omalizumab treatment.

Inclusion criteria

The inclusion criteria for children and young people participating in ADAPT were:

• they were aged 4–19 years

• they had severe eczema with (1) an objective SCORAD score (a validated eczema severity score) of over 40, which was (2) unresponsive to optimal topical therapy (potent topical steroids and/or topical calcineurin inhibitors) or systemic therapy, with (3) no impression of lack of compliance, (4) a (Children’s) Dermatology Life Quality Index [(C)DLQI] score of ≥ 10 and (5) where active skin infection had been ruled out and/or adequately treated

• they had a raised specific IgE (SpIgE) level (> 0.35 kUA/l) or SPT result (> 3 mm) to at least one food allergen or one aeroallergen and/or

• there was a clinical impression that allergic exposures caused worsening eczema

• they had a total IgE level of > 300 kU/l

• they had clinically proven IgE-mediated allergic disease including at least one of the following –

  • immediate hypersensitivity to a food as proven by raised SpIgE or SPT greater than the 95% positive predictive value or ≥ 8 mm, or a positive double-blind, placebo-controlled food challenge

  • allergic rhinoconjunctivitis as defined by sensitisation to a respiratory allergen and a clinical history of rhinoconjunctivitis symptoms when exposed to the relevant allergen

  • allergic asthma – a history of a cough, wheeze or shortness of breath that (1) was responsive to therapy with bronchodilators on two or more occasions in the previous 24 months, (2) required one visit to a physician in the previous 24 months and (3) occurred during the night, during early morning or on exercising in the intervals between exacerbations at any time in the previous 12 months, and (4) where allergic exacerbations can be clinically related to an allergen exposure with a corresponding positive SPT or SpIgE to the allergen

• they provided written informed consent to participate, or assent if appropriate.

Exclusion criteria

Children and young people were not able to participate if any of the following applied:

• They and/or their families were unable to comply with the regime of 2- to 4-weekly injections and clinic visits.

• There was evidence of underlying immune compromise, autoimmune disease or immune complex-mediated conditions.

• There was malignancy or a history of malignancy.

• There was a known cardiovascular or ischaemic cerebrovascular abnormality.

• There was other serious or uncontrolled systemic disease.

• The subject was pregnant or lactating.

• There was a known history of hypersensitivity or anaphylaxis to anti-IgE injections or its constituents.

• There was insufficient understanding of the trial assessments.

• They had participated in a clinical trial of an IMP in the previous 60 days or (if known) four half-lives of the medication under investigation, whichever was greater. In this case, entry may have been delayed until the appropriate time.

• The investigator felt that there was a good clinical reason why the child or young person was unsuitable for the study.

Informed consent

Please see editorial documentation (www.journalslibrary.nihr.ac.uk/programmes/eme/111424/#/; accessed July 2019).

Study flow chart

Figure 1 shows the study flow chart.

FIGURE 1.

Study flow chart.

Written informed consent was obtained from all participants before any study procedure was carried out. The participant (and/or parent/guardian), had the opportunity to review the participant information sheet and participant consent form prior to participation. Informed consent was taken by a suitably qualified and experienced medical doctor, as delegated by the chief investigator. Information sheets and assent forms for different age groups were available and verbal assent was obtained for all participants.

Randomisation and allocation procedure

A secure web-based randomisation system was used to allocate participants who fulfilled the eligibility criteria and consented to participate to receive either omalizumab or placebo in a 1 : 1 allocation ratio. The allocation sequence was computer generated by the UK Clinical Research Collaboration-registered King’s College London Clinical Trials Unit, designed in conjunction with an independent statistician. Participants were allocated to the arms using minimisation with the variables (1) IgE level (≤ 1500 or > 1500 kU/l) and (2) age (< 10 or ≥ 10 years).

Strict adherance to established procedures maintained separation between staff involved in outcome measurements and staff who delivered treatment. Study team physicians, researchers and research nurses involved in primary outcome assessments, participants and participants’ families were blinded to treatment allocation for the duration of the study. Randomisation details were electronically delivered to the independent pharmacy team, and the preparation and administration of the treatment was restricted to an allocated unblinded group of trained clinical staff to maintain blinding. Unblinded clinical staff collected and returned used vials of active or placebo medication to the pharmacy. Unblinded staff prepared and administered treatment in a closed treatment room with obscured glass, separate from the main clinical area. Blinded staff were not permitted entry to this area during the preparation and administration of the treatment. Thus, staff members who obtained outcome measurements were separated from any handling of the intervention, and unblinded clinical staff were not involved in trial-related primary outcome assessments.

All participants were given an emergency card with the contact details of the pharmacy department. This was carried for the duration of the trial and unblinding could be provided in clinically relevant situations to treating clinicians by the pharmacy department.

Treatment

Each participant was enrolled for 48 weeks, comprising 24 weeks of treatment followed by 24 weeks of follow-up.

The active treatment, omalizumab, and the placebo injection were manufactured by Novartis Pharmaceuticals UK Ltd and they were supplied as 150-mg single-use vials containing powder for reconstitution. The comparator placebo was formulated to be comparable in appearance and to contain the same excipients. The latest manufacturer’s dosing tables (according to the current Summary of Product Characteristics at the time of the study) were used to guide dose and dosing frequency of omalizumab and placebo. This was determined by baseline total IgE level (kU/l) and body weight (kg) at the randomisation visit. The weight measurement was repeated at baseline if the randomisation weight was on the borderline of two different doses. The dose of omalizumab that was closest to that child’s weight and IgE levels as stated on the dosing table was used. The dosing tables define doses for total IgE levels between 30 kU/l and 1500 kU/l. Participants with total IgE levels above 1500 kU/l received the maximum dose for their weight. Doses were 75 mg to 600 mg every 2 or 4 weeks, with an equivalent volume for placebo doses calculated in the same way. The dose remained unchanged over the 24 weeks of treatment. Subcutaneous administration of the active or placebo medication was undertaken in the deltoid region or thigh. Up to four injections were required at each visit. Local anaesthesia with topical local anaesthetic [lidocaine cream (LMX, Ferndale)] or chloroethane (Cryogesic, Accorus Therapeutics Ltd) spray was employed in accordance with participant preference.

Outcome measures

Pre visit 1 questionnaire

Patients were screened briefly in person or by telephone, using a pre visit 1 questionnaire. The study was explained to the patient/parent and any initial questions were addressed. The information sheets and consent forms, as well as any additional information requested by the families, such as the Xolair Summary of Product Characteristics, were then sent to the families by e-mail or post for them to consider further. If they were eligible after this initial screening, they were given the opportunity to be contacted by a staff member after having time to consider the information, or to call to book an appointment for a screening visit.

Screening visit

The purpose of the screening visit was to fully assess the child’s eligibility for the study. At the screening visit, the purpose of the visit and the study was explained once again, any outstanding queries were addressed and/or patients and parents or their guardians were asked to sign the relevant consent form(s).

A detailed history of the participant’s eczema, other atopic conditions and their medication, and general and family history were taken. The participants underwent a general examination and their eczema was assessed clinically using the SCORAD and EASI scoring systems. Questionnaires [(C)DLQI, POEM and PADQLQ] were completed. SPTs were carried out and blood was taken to assess their atopic status and as a general screen, including an assessment of vitamin D and iron deficiencies. Swabs were taken to record colonisation, as well as from the nose, throat and groin as a screen for meticillin-resistant Staphylococcus aureus (MRSA). MRSA-positive patients were treated and isolated from contact with other patients during their visits until three MRSA-negative swabs were obtained. Urinalysis was taken as a baseline screen and pregnancy testing was carried out on female patients who had attained menarche, as pregnant participants would not be eligible for participation as the safety of omalizumab in pregnancy had not been established.

Baseline and randomisation visit

Patients who were eligible at screening were invited to return for the baseline and randomisation visit and to start treatment.

Their eligibility criteria were rechecked to ensure that they were still eligible for the study. An examination to reassess their eczema by SCORAD and EASI was carried out, and their QoL questionnaires were repeated. Medical photographs were taken to document the extent of their eczema. They were then randomised using the online randomisation system. This alerted the pharmacist to dispense the active or placebo drug. The procedure for drug preparation and administration is outlined separately. The active/placebo drug was administered and the participant was observed for at least 2 hours. During the visit, TEWL measurements were also carried out in some patients, as this was approved during a later iteration of the protocol submitted with a substantial amendment after the first group of patients had been recruited. Participants who had provided consent to have skin biopsies carried out at baseline and at week 24 had their initial biopsy carried out at this visit.

Treatment visits

Participants attended for their treatment visits every 2 or 4 weeks during weeks 2–22, during which they were administered treatment in accordance with their assigned arms. The dose and frequency was determined by manufacturer’s dosing tables based on the participant’s weight and total IgE level. They would therefore attend visits at the predetermined frequency, regardless of whether they received the active or placebo drug. Visits were flexible within a period of ± 5 days (counted from the date of the first dose and regardless of the date of the previous visit).

All participants attended visits at least monthly and it was at these 4-weekly visits (weeks 4, 8, 16 and 20) that participants had their eczema assessed by the SCORAD and EASI scoring systems and at which they completed the questionnaires.

Week 24 visit

Patients were assessed for their primary outcome at this visit. They had a brief history taken and their eczema was clinically assessed using the SCORAD and EASI assessments. They completed the questionnaires and underwent a general physical examination. The SPTs were carried out and blood was taken, which included an assessment of vitamin D and iron deficiencies. TEWL measurements and urinalysis were carried out. The eczema status of each participant was documented once again by medical photography. Participants who had a skin biopsy taken at baseline had the option of a follow-up biopsy taken at this visit.

Week 36 and week 48 visits

Patients were assessed 3 and 6 months after completion of treatment to assess long-term benefits.

A general physical examination, SCORAD and EASI assessments, urinalysis and TEWL measurements were carried out. Patients also completed the questionnaires [(C)DLQI, POEM and PADQLQ].

Blood samples

Blood was taken in standard laboratory sample tubes [including EDTA (ethylenediaminetetraacetic acid), lithium heparin, clotted blood sample tubes, capillary tubes and sodium fluoride tubes] and sent to the in-house laboratory at Guy’s and St Thomas’ NHS Foundation Trust for full blood count, eosinophils, urea and electrolytes, liver function, vitamin D and iron levels and bone profile tests, and estimation of total and specific IgE levels. Further blood collected in citrate tubes was transported to the Paediatric Allergy Laboratory, part of the Asthma, Allergy and Lung Biology Department at King’s College London, based at Guy’s Hospital. There, the blood was separated into plasma and cells and stored at –80 °C in liquid nitrogen, and it was also used for genetic studies.

Skin swabs

Skin swabs were collected from the participants’ noses, throats and perinea for routine MRSA screening and from active areas of eczema. Clinically infected eczema was treated prior to baseline if required, to allow the participant to meet the inclusion criterion that ‘active skin infection had been ruled out and/or adequately treated’ to exclude infection as a cause for the severity of their eczema.

Skin biopsies

When consent was given, skin biopsies were taken at baseline from lesional and non-lesional skin, and a further biopsy was taken at 24 weeks. The sample was divided vertically into halves. One half was placed in RNAlater^® Solution (ThermoFisher Scientific, Waltham, MA, USA) and stored at 4 °C overnight, before being moved to –20 °C storage. The other half was frozen in Optimal Cutting Temperature (OCT) compound (Cryo-Embedding Medium O.C.T., TAAB Laboratories Equipment Ltd, Aldermaston, UK) and stored. Skin biopsy samples were stored in the Paediatric Allergy Laboratory, as were the blood samples.

Transepidermal water loss

Transepidermal water loss is a measure of the flux of water diffusing through the stratum corneum of the skin. TEWL was approved after a later protocol amendment; therefore, it was assessed in this study in only some patients, using the AquaFlux Model AF200 machine and version 9 of the AquaFlux software (AquaFlux, Blox Systems Ltd, London, UK). The AquaFlux uses the condenser-chamber measurement method. It is a non-invasive measurement whereby a probe with a closed chamber at its tip is placed against the skin for a few seconds while the reading is taken. The machine alerts the operator with an audible alarm when recording is successful. Three successful readings are taken from each site, and an average of the readings is taken. The sites examined at baseline are a lesional (the site of the skin biopsy if one is taken) and a non-lesional site. The sites examined at weeks 24, 36 and 48 are at or adjacent to the baseline lesional site.

Pregnancy testing

Testing for pregnancy was carried out on female participants who had reached menarche using urinary βhCG (beta-human chorionic gonadotropin) dipsticks on urine collected from the participant. This was carried out as it was unclear if the omalizumab has any effects in pregnancy or on the developing fetus, and pregnancy precluded participation in this study.

Recruitment

Participants were recruited from Guy’s and St Thomas’ NHS Foundation Trust and other hospitals. Guy’s and St Thomas’ NHS Foundation Trust has an in-house tertiary allergy and dermatology outpatient unit and wards, where the study team maintained a daily presence, as well as general paediatric services. The study employed a hub-and-spoke method of recruitment such that participants were identified from hospitals in and around London by participant identification centres (PICs), tertiary centres and secondary centres. Clinicians at PICs referred potential participants to the study team at Guy’s and St Thomas’ NHS Foundation Trust for assessment and recruitment to the study. The PICs identified participants mainly through outpatient clinic appointments, at which clinicians and nurses spoke directly to patients about the study and/or highlighted the opportunity to participate in ADAPT via posters in waiting rooms. Permission was sought to pass on the patient’s contact details to the ADAPT Study Team at Guy’s and St Thomas’ NHS Foundation Trust. Treatment was delivered at a single centre (Guy’s and St Thomas’ NHS Foundation Trust).

The study team developed strong links with dermatology teams at local hospitals. The study was highlighted at departmental meetings at local hospitals, general practitioners’ (GPs’) meetings, national and international dermatology conferences and to members of the National Eczema Society (NES), a patient support group.

The study team participated in activities such as a local fundraising day marking the 10-year anniversary of the host institution, the Evelina Children’s Hospital, and participant-orientated research awareness events, such as the International Clinical Trials Day and an art workshop for research participants.

Regular newsletters were sent to health-care professionals, participants and their families. Posters and lanyard-sized cards were also distributed. Information about the study was also provided in collaboration with the NES on its website and media outlets. The study hosted and maintained its own website with up-to-date information for children and young people and their families.

Retention of participants

The study team maintained an approachable relationship with participants and provided contact via telephone and e-mail to discuss any aspect of the study, distributing study newsletters and festive greetings during the course of the study. Participants were invited to research awareness events at the hospital – the International Clinical Trials Day celebrations included mock clinical trials, arts and crafts, entertainment, information about research taking place at the hospital and an awards ceremony to celebrate the contribution of research participants. 79 Families also took part in an art workshop, in collaboration with deadcatdreaming (www.deadcatdreaming.co.uk), entitled ‘The BIGGER Picture consultation’, to showcase their stories of taking part in research. 80

Trial Steering Committee: patient and public involvement member

Our PPI representative on the TSC was the training co-ordinator for the NES, a patient support group, and had personal experience of a family living with severe eczema. She was involved in discussions related to the design of the study. She facilitated links with NES, which communicated details of the study on its website and in its quarterly members’ magazine, Exchange. She was invited to all TSC meetings, including the results unblinding meeting. Her opinion was sought at the meeting regarding the outcome of the study. She felt that families are wary of topical steroids and that another treatment option would be welcomed by families, as long as they were kept fully informed during the decision-making process. She also reviewed the Plain English summary of this report.

Focus group interviews: design and development of the protocol

During the design phase of the study, a group of nine families of children and young people with eczema were interviewed to assess interest for a study of this kind and to critique the study design and participant information leaflets.

These families generally felt that more research was needed in the field of eczema, especially as current therapies were not meeting patient needs. They felt that there was little understanding about the disease from non-specialists who sometimes dismissed eczema as ‘minor and unimportant’, and that advice they were given in the community was ‘hit and miss’.

Families felt that the study was well designed and that the follow-up period was important to assess the long-term effects of the treatment. They liked the idea of a control and felt that it was important that the active and placebo arms were well matched. They agreed that it was important to address QoL issues, and felt that the QoL questionnaires chosen for the study covered many important aspects of the disease. Other families felt that the monitoring of other comorbidities with the PADQLQ questionnaire was a useful adjunct and would add much-needed information to the study.

They thought that clear written information was important. They found the participant information leaflets easy to read and liked the option of different assent forms for children and young people, tailored to the different age groups.

Families wanted to know more about safety and side effects of the drug, and felt that this should be well explained at the start. The study team ensured that as much time as needed would be spent with each family to discuss these issues before they were enrolled, and also submitted a REC amendment so that the Summary of Product Characteristics for omalizumab, which lists the side effects, could be shared with families if requested.

Practically, families highlighted issues with travel and the associated costs of travel, including the London Congestion Charge, an additional travel cost for vehicles that operates during peak hours in central London. There was a budget set aside to cover travel expenses. Other families thought that the timing of visits should be tailored to suit family and school schedules. In response to this, staff were scheduled to work flexibly with early start times or late finishes, should families prefer an early-morning or an evening appointment to avoid missing school. Other families felt that everyday life may get in the way of the follow-up visits 3 and 6 months after the end of the treatment period. They suggested that a good explanation at the start of the study would help to limit this loss to follow-up.

Families felt that it was crucial that the results of the study would be widely disseminated, and wanted the drug to be made available to other children and young people if it was deemed successful.

General statistical principles

Subgroup-blind analysis (i.e. as A vs. B) was conducted in accordance with the statistician analysis plan (www.journalslibrary.nihr.ac.uk/programmes/eme/111424/#/; accessed July 2019), which was finalised prior to database lock. Analysis was undertaken by the statistician who was subgroup blind (SC) and was based on the intention-to-treat (ITT) principle, that is, participants were analysed in the arm to which they were randomised regardless of subsequent treatment received. All regression analyses included the minimisation variables IgE level (≤ 1500 or > 1500 kU/l) and age (< 10 or ≥ 10 years) as covariates. This is because adjustment for stratification factors in the randomisation process maintains the correct type 1 error rates. In addition, for continuous outcomes, the outcome measured at baseline was included in regression analysis to increase power. Estimates are presented with 95% confidence intervals (CIs) and p-values. All statistical analyses were conducted using Stata^® version 15.1 (StataCorp LP, College Station, TX, USA).

Descriptive analysis

A Consolidated Standards of Reporting Trials (CONSORT) flow chart83 was constructed to summarise the flow of participants through the study (see Figure 2). Baseline characteristics were summarised by randomised arm to examine balance between the arms at baseline.

All outcomes were summarised by time point and treatment arm. The proportion of participants lost to follow-up and missing objective SCORAD values (primary outcome) was summarised by treatment arm and at each time point. The baseline characteristics age, sex, objective and total SCORAD, body mass index, asthma (yes/no), food allergy (yes/no), rhinoconjunctivitis (yes/no) and referral source (self-referred/tertiary) of those missing follow-up data were compared with the characteristics of those with complete follow-up data.

Treatment adherence, reasons for withdrawal and use of AST, rescue medication with oral prednisolone and potent topical steroids were also summarised by treatment arm.

Analysis of the primary outcome

A linear mixed model including observations at 8, 12, 16, 20 and 24 weeks was used to obtain an estimate of the mean treatment arm difference in objective SCORAD scores at 24 weeks. The model included fixed effects for time, time*treatment arm interaction, baseline objective SCORAD score, IgE level (≤ 1500 or > 1500 kU/l) and age (< 10 or ≥ 10 years). To allow for between-participant differences, the model included a random intercept at the participant level. An unstructured covariance matrix was chosen to model the covariance structure because it allows for all variances and covariances to be distinct. 84 In keeping with the ITT principle, all participants who provided data from at least one follow-up visit (at 8, 12, 16, 20 or 24 weeks) were included in the analysis as randomised. All missing response values were assumed to be missing at random (MAR) (i.e. the probability that the response is missing does not depend on the value of the response after controlling for the observed variables). The results of the primary outcome analysis were verified by an independent statistician.

Planned sensitivity analyses for the primary outcome were conducted. These included:

• Adjustment for initiation of AST (within the primary analysis model).

• Adjustment for initiation of AST and rescue medication with oral prednisolone (within the primary analysis model).

• Adjustment for initiation of AST, rescue medication with oral prednisolone and potent topical steroids (within the primary analysis model).

• Use of multiple imputation (MI) to explore the impact of a worse outcome post initiation of AST. The primary analysis model was retained for use in the sensitivity analysis, following MI.

• Use of MI to explore the impact of a worse outcome for participants with missing outcome data. The primary analysis model was retained for use in the sensitivity analysis, following MI.

• An adherence-adjusted analysis. The complier-average causal effect (CACE) was estimated using a two-stage least squares instrumental variable regression for the primary end point. Here, we defined ‘compliers’ as those who complete more than 50% of injections (i.e. injections received relative to injections planned for the 24-week study period in both arms of the study). Randomisation was used as an instrumental variable for treatment received, with the same covariates as in primary analysis models.

Analysis of secondary outcomes

Secondary outcome analysis focused on the outcome at week 24. Linear regression models were used for the continuous secondary outcomes (total SCORAD, EASI, POEM, CDLQI and PADQLQ). Binary outcomes (treatment failure and AST use) were analysed using logistic regression models. Zero-inflated Poisson regression models were used to analyse counts (infective episodes of eczema, eczema exacerbations and the number of positive SPTs at 24 weeks).

Adverse events were tabulated separately by type (adverse events, adverse reactions, unexpected adverse reactions, SAEs, serious adverse reactions or unexpected serious adverse reactions). Poisson regression models were used to estimate relative risks, risk differences and incidence rate ratios (IRRs) of non-serious events by body system class. A volcano plot, which plotted the risk difference of the non-serious adverse events by MedDRA-preferred term between the treatment arms against the p-value from a Fisher’s exact test, was examined to identify the events with the strongest evidence for between-arm differences.

Subgroup analysis

A subgroup analysis was planned a priori to investigate whether or not the treatment effect differed by adherence to treatment. Adherence was defined as the injections received relative to the injections planned for the 24-week study period (≤ 50 or > 50%, ≤ 75 or > 75%, ≤ 90 or > 90%).

Exploratory analysis

A longitudinal analysis, using a linear mixed model, was undertaken to determine the difference in objective SCORAD score at 48 weeks. The analysis model was the same as in the primary analysis but included additional data observations at 36 and 48 weeks.

Post hoc analyses

Because the dose of omalizumab depended on total IgE levels, but was capped at an IgE level of 1500 kU/l, and the baseline total IgE levels were higher than expected, a post hoc analysis was conducted to explore the interaction between baseline total IgE level and treatment arm. This was conducted by adding an interaction term between treatment arm, baseline total IgE level (as a continuous variable) and time to the primary analysis model.

The observed treatment effect on eczema QoL was small and not significant when measured using the POEM but was larger and statistically significant when measured using the (C)DLQI. Following concerns that there may have been differences in how the POEM was completed by younger children, a post hoc analysis investigated whether or not the treatment effect on POEM differed by age group (< 10 or ≥ 10 years). The linear regression model used to analyse the week 24 POEM scores was extended to include the age group*treatment arm interaction.

Statistical software

All analyses were carried out using Stata version 15.1.

Primary analysis

The unadjusted mean objective SCORAD improved in both treatment arms during the 24-week treatment period (Table 8 and Figure 3). The improvement was significantly greater in the omalizumab arm. After adjustment for baseline objective SCORAD score, age and IgE level within a linear mixed model, the mean treatment arm difference (i.e. the difference between the change in objective SCORAD in the omalizumab arm vs. the change in the placebo arm) at week 24 was –6.9 (95% CI –12.2 to –1.5; p = 0.013) (Figure 4). Although the average treatment effect was smaller than what we aimed to detect (–13.5), the 95% CI includes the MCID of 8.2 reported by Schram et al. 72 and the MCID of 8.5 that was calculated using ADAPT data, suggesting that an important average clinical benefit cannot be ruled out.

FIGURE 3.

Mean objective SCORAD score over time by treatment arm. Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

FIGURE 4.

Adjusted treatment arm difference in objective SCORAD score from primary analysis during the treatment phase.

Within each arm, the unadjusted mean objective SCORAD score improved by –12.4 in the omalizumab arm [from 55.5 (SD 9.5) at baseline to 43.1 (SD 12.5) at 24 weeks] and by –5.1 in the placebo arm [from 54.3 (SD 7.7) at baseline to 49.2 (SD 11.3) at 24 weeks]. This difference at 24 weeks exceeded the MCID in the omalizumab arm.

Post hoc analysis for primary outcome by baseline immunoglobulin E level

The majority of participants had very high total IgE levels. However, further analyses suggested that participants with lower baseline total IgE levels may respond more favourably to treatment.

Figure 5 shows the unadjusted relationship between baseline total IgE level and 24-week objective SCORAD score, with line of linear best fit by treatment arm.

FIGURE 5.

Scatterplot of baseline total IgE level against week 24 objective SCORAD score.

To investigate whether or not the treatment effect differed by baseline total IgE level, the primary analysis model was first fitted using baseline total IgE level (kU/l) in place of categorised baseline total IgE level (< 1500 or ≥ 1500 kU/l). The model was then extended to include the baseline total IgE level by treatment*time interaction (kU/l). The study was not powered to detect such an interaction and, given the small sample size, this test had limited power. The interaction effect at week 24, representing the average increase in objective SCORAD at week 24 for a 1000-unit increase of IgE at baseline for participants in the omalizumab arm, was 0.3 (95% CI –0.1 to 0.7; p = 0.193).

Although the interaction effect was insignificant, after allowing for a differential treatment effect by baseline IgE level there was some evidence of variation in treatment effect by baseline IgE level (Figure 6). After adjustment for the baseline total IgE level*treatment interaction, the treatment effect for a median baseline IgE level of 8373 kU/l was –7.9 (95% CI –13.7 to –2.2; p = 0.007). The treatment effect for a baseline IgE level of 4556 kU/l (25th percentile) was –9.0 (95% CI –15.4 to –2.5; p = 0.006) and the treatment effect for a baseline IgE level of 18,506 kU/l (75th percentile) was –5.2 (95% CI –11.0 to 0.7; p = 0.083). The evidence suggested a stronger treatment effect for lower baseline IgE levels. However, wider CIs for the estimated treatment effect at higher baseline IgE levels (see Figure 6) indicate that more data are required to reach definitive conclusions for individuals with very elevated IgE levels.

FIGURE 6.

Treatment effect (95% CI) on objective SCORAD by baseline total IgE level using a mixed model.

Eczema severity

Eczema severity was also assessed by the total SCORAD score (the sum of the objective and subjective SCORAD scores) and the EASI score. Both scores showed significant treatment arm differences in favour of omalizumab. This was not reflected by the POEM score. There was, however, a greater treatment effect in the POEM score among younger participants. This is detailed in section Patient-Oriented Eczema Measure below.

Total SCORAD

Mean total SCORAD scores by visit and treatment arm are shown in Figure 7. The treatment arm difference (i.e. the difference between the change in total SCORAD in the omalizumab arm vs. the change in the placebo arm) at week 24, adjusted for baseline total SCORAD score, age and IgE level, was –8.3 (95% CI –15.1 to –1.1; p = 0.024). The MCID was 8.7 (Table 14).

FIGURE 7.

Mean total SCORAD scores over time by treatment arm. Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

TABLE 14 - Mean total SCORAD scores over time by treatment arm

Time point	Total SCORAD score by treatment arm				Total (n)	Unadjusted mean difference (95% CI)	Adjusted mean difference (95% CI)	p-value
	Omalizumab		Placebo
	n	Mean (SD)	n	Mean (SD)
Baseline	30	69.5 (10.7)	32	69.1 (9.2)	62	–	–
Week 4	30	64.7 (13.9)	30	66.2 (14.2)	60	–1.5 (–8.7 to 5.8)	–
Week 8	29	58.2 (13.5)	28	65.0 (12.5)	57	–6.8 (–13.7 to 0.1)	–
Week 12	29	60.9 (12.9)	28	59.5 (12.4)	57	1.4 (–5.3 to 8.1)	–
Week 16	29	57.0 (17.0)	28	56.9 (13.6)	57	0.1 (–8.1 to 8.3)	–
Week 20	29	56.6 (12.5)	28	61.0 (14.9)	57	–4.4 (–11.6 to 2.9)	–
Week 24	30	53.1 (15.4)	30	60.9 (13.5)	60	–7.9 (–15.4 to –0.4)	–8.3 (15.5 to –1.1)	0.024
Week 36	30	55.7 (18.7)	31	59.1 (14.2)	61	–3.3 (–11.8 to 5.2)	–
Week 48	30	56.6 (16.4)	31	60.1 (18.9)	61	–3.6 (–12.6 to 5.5)	–

Within each arm, the mean total SCORAD score improved by –16.4 [from 69.5 (SD 10.7) at baseline to 53.1 (SD 15.4) at week 24] in the omalizumab arm and by –8.2 [from 69.1 (SD 9.2) at baseline to 60.9 (SD 13.5) at week 24] in the placebo arm. This exceeded the MCID for the omalizumab arm.

Eczema Area and Severity Index

Figure 8 shows the mean EASI scores by visit and treatment arm. A statistically significant and clinically important difference in EASI score by treatment arm was identified at week 24. The treatment arm difference (i.e. the difference between the change in EASI score in the omalizumab arm vs. the change in the placebo arm) at week 24 after adjustment for baseline EASI score, age and IgE level was –6.7 (95% CI –13.2 to –0.1; p = 0.046) (Table 15), compared with an MCID of 6.6.

FIGURE 8.

Mean EASI scores over time by treatment arm. Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

TABLE 15 - Mean EASI scores over time by treatment arm

Time point	EASI score by treatment arm				Total (n)	Unadjusted mean difference (95% CI)	Adjusted mean difference (95% CI)	p-value
	Omalizumab		Placebo
	n	Mean (SD)	n	Mean (SD)
Baseline	30	45.5 (10.1)	32	43.5 (11.1)	62	–	–	–
Week 4	30	40.9 (12.5)	30	41.1 (12.2)	60	–0.2 (–6.6 to 6.2)	–	–
Week 8	29	38.8 (12.6)	28	43.3 (12.8)	57	–4.5 (–11.3 to 2.2)	–	–
Week 12	29	40.2 (11.4)	28	40.3 (12.2)	57	–0.1 (–6.4 to 6.1)	–	–
Week 16	29	37.4 (13.5)	28	35.8 (14.4)	57	1.6 (–5.8 to 9.0)	–	–
Week 20	29	36.9 (9.8)	28	37.4 (14.7)	57	–0.4 (–7.0 to 6.2)	–	–
Week 24	30	32.8 (13.5)	30	38.4 (12.4)	60	–5.6 (–12.3 to 1.1)	–6.7 (–13.2 to –0.1)	0.046
Week 36	30	35.1 (16.0)	31	37.6 (14.5)	61	–2.5 (–10.3 to 5.3)	–	–
Week 48	30	34.9 (15.1)	31	37.4 (15.6)	61	–2.6 (–10.4 to 5.3)	–	–

Within each arm, the mean EASI score improved by –12.7 [from 45.5 (SD 10.1) at baseline to 32.8 (SD 13.5) at week 24] in the omalizumab arm and by –5.1 [from 43.5 (SD 11.1) at baseline to 38.4 (SD 12.4) at week 24] in the placebo arm. This exceeded the MCID for the omalizumab arm.

Patient-Oriented Eczema Measure

The treatment arm difference in the week 24 POEM scores, after adjustment for baseline POEM score, age and IgE level for omalizumab versus placebo was –1.1 (95% CI –4.6 to 2.4; p = 0.527) (Table 16). The average treatment effect was much lower than the MCID of 3.4, and the 95% CI included the possibility of no difference between the arms or a favouring of placebo. Mean POEM scores by visit and treatment arm are shown in Figure 9.

TABLE 16 - Mean POEM scores over time by treatment arm

Time point	POEM score by treatment arm				Total (n)	Unadjusted mean difference (95% CI)	Adjusted mean difference (95% CI)	p-value
	Omalizumab		Placebo
	n	Mean (SD)	n	Mean (SD)
Baseline	30	20.7 (4.6)	32	22.2 (3.9)	62	–	–	–
Week 4	30	17.5 (6.3)	30	19.4 (5.7)	60	–1.9 (–5.0 to 1.2)	–	–
Week 8	29	14.7 (6.7)	28	17.5 (6.6)	57	–2.8 (–6.3 to 0.7)	–	–
Week 12	29	15.3 (6.6)	28	16.6 (7.0)	57	–1.3 (–5.0 to 2.3)	–	–
Week 16	29	14.7 (6.0)	28	15.0 (7.9)	57	–0.3 (–4.0 to 3.4)	–	–
Week 20	29	14.3 (7.5)	28	15.3 (7.5)	57	–0.9 (–4.9 to 3.0)	–	–
Week 24	30	14.0 (7.5)	30	16.0 (6.7)	60	–2.0 (–5.7 to 1.7)	–1.1 (–4.6 to 2.4)	0.527
Week 36	30	15.0 (7.8)	31	15.6 (7.1)	61	–0.6 (–4.4 to 3.3)	–	–
Week 48	30	15.3 (8.0)	31	14.8 (8.1)	61	0.5 (–3.7 to 4.6)	–	–

FIGURE 9.

Mean POEM scores over time by treatment arm. Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Post hoc analysis for Patient-Oriented Eczema Measure scores by baseline age

In contrast to the eczema severity scores and the eczema QoL result for the (C)DLQI, the observed treatment arm difference for the POEM was much smaller than the reported MCID for POEM (MCID = 3.4). 72 A post hoc analysis investigated whether or not the treatment effect on POEM differed by age group (< 10 or ≥ 10 years), as it was noted that in younger children the questionnaire was often completed by the carer but in older participants it was completed by the child themselves. There was a significant interaction between age (< 10 or ≥ 10 years) and treatment arm on POEM score at 24 weeks, suggesting that the treatment effect differs by age group. The interaction effect, representing the average difference in POEM score at week 24 for a participant aged ≥ 10 years in the omalizumab arm, was 8.0 (95% CI 1.5 to 14.6; p = 0.017). After adjustment for baseline POEM score, age, total IgE level and the age*treatment interaction, a larger treatment effect in favour of omalizumab with POEM was observed for children aged < 10 years (–5.2, 95% CI –10.0 to –0.5; p = 0.031). The treatment effect in children and young people aged ≥ 10 years was 2.8 (95% CI –1.8 to 7.4; p = 0.230) in favour of placebo (Figure 10).

FIGURE 10.

Treatment effect (95% CI) on week 24 POEM score by baseline age group (< 10 or ≥ 10 years).

Potent topical steroid cream usage

Participants in the placebo arm used more potent topical steroids than participants in the omalizumab-treated arm. The median number of days of use of potent topical steroids in the placebo arm was 48% higher than that in the active arm at week 24 and 55% higher at week 48. Participants in the placebo arm also used potent topical steroids on twice the amount of BSA at 24 weeks and 1.7 times BSA at 48 weeks compared with omalizumab-treated participants. The median weight of potent topical steroids used (grams) was 76% higher in the placebo arm than in the omalizumab arm at weeks 24 and 48.

Week 24 results

Number of days of potent topical steroid use at week 24

During the 24 week follow-up period, 589 records of potent topical steroid use were collected from 58 participants: 27 in the omalizumab arm and 31 in the placebo arm. For each participant, the number of days of potent topical steroid use over the 24-week (168-day) follow-up period was calculated (Figures 11 and 13). The median number of days of potent topical steroid use was 109 in the omalizumab arm [IQR 34–164 days, minimum = 0 (n = 3) and maximum = 170 days]. In the placebo arm, the median number of days of potent topical steroid use was 161 [IQR 82–171 days, minimum = 0 (n = 1) and maximum = 191 days]. The between treatment arm difference was marginally insignificant (Mann Whitney U-test, p = 0.067). The visit windows allowed for a 2-week leeway either side of 24 weeks, thus some patients had a 24-week visit that fell just outside the 24-week time frame and required potent topical steroids for these additional days.

FIGURE 11.

Number of days of potent topical steroid use by treatment arm during 24-week treatment/follow-up period. (a) Omalizumab; and (b) placebo.

Body surface area covered and weight of potent topical steroid used at week 24

Further details of the usage of potent topical steroids over the 24-week treatment period are given in Table 17. For each participant who had more than one record of topical steroid usage, we calculated the average percentage of BSA, strength, frequency per week and frequency per day of application over the 24-week follow-up period prior to summarising by treatment arm. Participants in the omalizumab arm used potent topical steroids over a median of 15.5% of their BSA, compared with 31.3% BSA in the placebo arm. Participants in the omalizumab arm used a median weight of 58 g, compared with 102 g in the placebo arm, over the 24-week period.

TABLE 17 - Details of potent topical steroid use over the 24-week treatment/follow-up period

%BSA, percentage of BSA.

Denominator includes the number of participants followed up for the full 24-week follow-up period. It excludes one participant in the placebo arm who withdrew from treatment following week 6 owing to transportation issues.

Data on weight used were collected for only 408 (69.3%) potent topical steroid records [174 (69.3%) omalizumab; 234 (69.2%) placebo].

Data on frequency per week and day available for only n = 26 in the omalizumab arm.

Note

There were an additional 53 potent topical steroid records in the database that did not include an exact start and stop date, and therefore could not be not included in the potent topical steroid use figures (21 from the omalizumab arm in 12 participants; 32 from the placebo arm in 15 participants).

Potent topical steroid use	Treatment arm
	Omalizumab	Placebo
Number of recorded uses of potent topical steroids	251	338
Number of participants using potent topical steroids,a n (%)	27 (90)	31 (100)
Number of days of use per participant
Median (IQR)	109 (34–164)	161 (82–171)
Minimum, maximum	0, 170	0, 191
If used, average %BSA per participant
Median (IQR)	15.5 (9.9–46.3)	31.3 (14.0–55.0)
Minimum, maximum	0.0, 95.0	1.0, 96.3
If used, total weight (g) per participantb
Median (IQR)	58 (14–125)	102 (55–209)
Minimum, maximum	0, 489	5, 307
If used, average frequency per week per participantc
Median (IQR)	5 (3–6)	4 (3–6)
Minimum, maximum	2, 7	1, 7
If used, average frequency per day per participantc
Median (IQR)	1 (1–2)	1 (2–2)
Minimum, maximum	1, 2	1, 2

Week 48 results

Number of days of potent topical steroid use at week 48

Over the full 48-week follow-up period, 703 records of potent topical steroid use were collected from 61 participants: 30 in the omalizumab arm and 31 in the placebo arm. For each participant, the number of days of topical steroid use over the 48-week (336-day) follow-up period was calculated (Figures 12 and 13). The median number of days of potent topical steroid use was 188 in the omalizumab arm (IQR 49 to 299 days, minimum = 7 and maximum = 362 days). In the placebo arm, the median number of days of potent topical steroid use was 291 (IQR 111 to 336 days, minimum = 3 and maximum = 369 days).

FIGURE 12.

Number of days of potent topical steroid use by treatment arm during the 48-week follow-up period. (a) Omalizumab; and (b) placebo.

FIGURE 13.

Proportion of participants using potent topical steroids over the 48-week follow-up period. Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Body surface area and weight used at week 48

Data on the usage of potent topical steroids over the full 48-week follow-up period are summarised in Table 18. Participants in the omalizumab arm used potent topical steroids over a median of 18.2% of their BSA, compared with 32.2% BSA in the placebo arm.

TABLE 18 - Details of potent topical steroid use over the 48-week follow-up period

%BSA, percentage of BSA.

Denominator includes the number of participants followed up for the full 48-week follow-up period. It excludes one participant in the placebo arm, who withdrew from treatment following week 6 owing to transportation issues.

Data available for only n = 29 in the omalizumab arm.

Data available for only n = 30 in the placebo arm.

Data on weight used were collected for only 461 (65.6%) potent topical steroid records [203 (66.6%) omalizumab; 258 (64.8%) placebo].

Note

There were an additional 53 potent topical steroid records in the database that did not include an exact start and stop date, and therefore could not be not included in the topical steroid use figures (21 from the omalizumab arm in 12 participants and 32 from the placebo arm in 15 participants).

Potent topical steroid use	Treatment arm
	Omalizumab	Placebo
Number of recorded uses of potent topical steroids	305	398
Number of participants using potent topical steroids,a n (%)	30 (100)	31 (100)
Number of days of use
Median (IQR)	188 (49–299)	291 (111–336)
Minimum, maximum	7, 362	3, 369
Average %BSAb per participantc
Median (IQR)	18.2 (12.6–46.3)	32.2 (15.1–51.6)
Minimum, maximum	0.0, 74.9	1.0, 86.4
Total weight (g) per participantd
Median (IQR)	82 (27–181)	144 (65–260)
Minimum, maximum	0, 1022	5, 452
Average frequency per week per participantb
Median (IQR)	5 (3–6)	4 (3–6)
Minimum, maximum	2, 7	1, 7
Average frequency per day per participantb
Median (IQR)	1 (1–2)	1 (1–2)
Minimum, maximum	1, 2	1, 2

Out of the 703 records of potent topical steroid usage, 461 (65.6%) collected information on weight used (g). The median total weight used over 48 weeks was 82 g (IQR 27–181 g) in the omalizumab arm (n = 25) and 144 g (IQR 65–260 g) in the placebo arm (n = 28). Because not every record of use that collected the median total weight used over the 48-week period in each arm is available, this is an underestimate of the true value and therefore should be interpreted with caution.

Treatment failure

More participants in the placebo arm were defined as treatment failures, but the numbers were small, as outlined in this section.

Treatment failure was defined as participants who received two courses of rescue therapy with oral prednisolone between 12 and 24 weeks of treatment. One participant was a treatment failure in the omalizumab arm versus three in the placebo arm (3.3% vs. 9.7%, respectively). A logistic regression model was used to adjust the treatment arm difference for baseline age and total IgE level. The adjusted odds ratio (OR) of treatment failure between the treatment arms was 0.3 (95% CI 0.03 to 3.16; p = 0.319) (Table 19). As the numbers were small, there is a high level of uncertainty in the treatment effect estimate.

Alternative systemic therapy (i.e. systemic immunosuppression)

More participants in the placebo arm than in the omalizumab arm went on to receive alternative systemic eczema therapies, but the numbers were small, as outlined in this section.

Alternative systemic therapy (including azathioprine, ciclosporin and methotrexate) was defined for participants in whom ‘a) alternative systemic therapy has been started as a result of treatment failure as defined in the above section or b) where alternative systemic therapy is started after 12 weeks and by 30 weeks’. 1

A total of five participants started AST within 30 weeks (see Table 9 and Table 20): one in the omalizumab arm and four in the placebo arm (3.3% vs. 12.9%, respectively). The one participant in the omalizumab arm was first initiated on methotrexate just after 9 weeks and again at 14 weeks. This participant had been withdrawn from omalizumab treatment by study investigators and the DMEC (who remained blinded to treatment allocation) after receiving their first three injections over concerns of a potential adverse reaction to the study IMP/placebo (see Table 4). In the placebo arm, one participant was initiated on ciclosporin at the week 24 visit, two participants were initiated on methotrexate (one just after week 24 and one just after week 29) and one further participant was initiated on azathioprine just after 26 weeks. A logistic regression model was used to adjust the treatment arm difference for baseline age and total IgE level, and treatment arm differences were observed (adjusted OR 0.2, 95% CI 0.02 to 2.13; p = 0.192).

Seven patients in the omalizumab arm started the study while on systemic immunosuppression or UV therapy, with one patient on both ciclosporin and oral steroids. All of these participants continued these treatments to the primary end point at week 24. As described previously, one additional participant started systemic immunosuppression by week 30. Thus, a total of eight patients in the omalizumab arm were on systemic immunosuppression or UV therapy by the predefined week 30 end point, one of whom started this treatment during the predefined period.

Four patients in the placebo arm started the study on systemic immunosuppression or UV therapy. As described previously, none of these stopped systemic immunosuppression during the 24-week treatment period. No placebo patients started treatment with systemic immunosuppression prior to week 24. Four additional patients started treatment with systemic immunosuppression between week 24 and by the predefined week 30. Thus, a total of eight patients in the placebo arm were on systemic immunosuppression by week 30, four of whom started treatment during the course of the predefined period.

As instructed, no participants stopped systemic immunosuppression or UV therapy during the course of treatment. Our definition of the need for systemic immunosuppression or UV therapy by week 30 therefore encompassed only those additional participants who started these treatments during the course of the study. There was one participant in the omalizumab arm and four participants in the placebo arm who met this criterion. Although more participants in the placebo arm went on to receive alternative systemic eczema therapies, as the numbers were small there was a level of uncertainty in the treatment effect estimate. By 48 weeks, six participants in each arm were initiated on systemic immunosuppression or UV therapy (see Table 9).

One of the participants in the omalizumab arm and two of the participants in the placebo arm who started AST were also treatment failures. Thus, if we consider the treatment burden as treatment failure (as described previously) or initiation on AST after 12 weeks and by 30 weeks, there was one failure in the omalizumab arm (3.3%) and five in the placebo arm (16.1%).

Quality of life

There was a clinically and statistically significant treatment arm difference in (C)DLQI scores, with a halving of the score from baseline to week 24. There was also a clinically significant treatment arm difference in the PADQLQ score.

(Children’s) Dermatology Life Quality Index

Mean (C)DLQI scores by visit and treatment arm are shown in Figure 14. The treatment arm difference in the week 24 (C)DLQI score after adjustment for baseline (C)DLQI score, age and IgE level for omalizumab versus placebo was –3.5 (95% CI –6.4 to –0.5; p = 0.022) (Table 21). The point estimate was greater than the MCID of 3.3 for the DLQI. The mean (C)DLQI score in the omalizumab arm halved from 17 (SD 5.6) at baseline to 8.5 (SD 5.9) at week 24.

FIGURE 14.

Mean (C)DLQI scores over time by treatment arm. Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

TABLE 21 - Mean (C)DLQI scores over time by treatment arm

Time point	(C)DLQI scores by treatment arm				Total (n)	Unadjusted mean difference (95% CI)	Adjusted mean difference (95% CI)	p-value
	Omalizumab		Placebo
	n	Mean (SD)	n	Mean (SD)
Baseline	30	17.0 (5.6)	32	17.0 (4.5)	62	–	–	–
Week 4	30	13.3 (6.2)	30	15.2 (5.8)	60	–1.9 (–5.0 to 1.1)	–	–
Week 8	29	10.0 (6.5)	28	12.9 (5.0)	57	–2.9 (–6.0 to 0.2)	–	–
Week 12	29	9.7 (5.6)	28	12.0 (4.9)	57	–2.3 (–5.2 to 0.5)	–	–
Week 16	29	8.8 (5.9)	28	10.8 (5.7)	57	–2.0 (–5.1 to 1.1)	–	–
Week 20	29	8.3 (5.8)	28	10.6 (6.0)	57	–2.3 (–5.5 to 0.8)	–	–
Week 24	30	8.5 (5.9)	30	11.8 (5.6)	60	–3.3 (–6.3 to –0.3)	–3.5 (–6.4 to –0.5)	0.022
Week 36	30	8.7 (6.2)	31	9.9 (5.7)	61	–1.2 (–4.3 to 1.8)	–	–
Week 48	30	8.9 (7.0)	31	12.0 (7.3)	61	–3.0 (–6.7 to 0.6)	–	–

Paediatric Allergic Disease Quality of Life Questionnaire

There was a clinically significant effect on the PADQLQ, which is discussed in this section.

Mean PADQLQ scores by visit and treatment arm are shown in Figure 15. The treatment arm difference in the week 24 PADQLQ after adjustment for baseline PADQLQ score, age and IgE level for omalizumab versus placebo was –0.5 (95% CI –0.9 to 0.0; p = 0.050) (Table 22), indicating a clinically important difference by treatment arm where the MCID is 0.33.

FIGURE 15.

Mean PADQLQ scores over time by treatment arm. Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

TABLE 22 - Mean PADQLQ scores over time by treatment arm

Time point	PADQLQ scores by treatment arm				Total (n)	Unadjusted mean difference (95% CI)	Adjusted mean difference (95% CI)	p-value
	Omalizumab		Placebo
	n	Mean (SD)	n	Mean (SD)
Baseline	29	2.6 (1.3)	32	2.9 (1.3)	61	–	–	–
Week 4	30	2.1 (1.4)	30	2.5 (1.2)	60	–0.5 (–1.1 to 0.2)	–	–
Week 8	29	1.6 (1.2)	28	2.3 (1.2)	57	–0.7 (–1.3 to –0.1)	–	–
Week 12	29	1.5 (1.1)	28	2.1 (1.2)	57	–0.6 (–1.2 to 0.0)	–	–
Week 16	29	1.3 (1.0)	28	1.9 (1.3)	57	–0.5 (–1.1 to 0.1)	–	–
Week 20	28	1.2 (1.0)	28	1.9 (1.3)	56	–0.7 (–1.3 to 0.0)	–	–
Week 24	30	1.3 (1.0)	30	1.9 (1.1)	60	–0.6 (–1.1 to –0.1)	–0.5 (–0.9 to 0.0)	0.050
Week 36	30	1.4 (0.9)	31	1.8 (1.3)	61	–0.5 (–1.0 to 0.1)	–	–
Week 48	30	1.5 (1.2)	30	1.9 (1.3)	60	–0.4 (–1.0 to 0.3)	–	–

Assessment of allergic markers

There was a general reduction in the number of positive SPTs in the omalizumab arm compared with the placebo arm. There was a reduction in the median total IgE level in the omalizumab arm and an increase in the placebo arm. There is no strong evidence of a reduction in SpIgE levels with omalizumab.

Skin prick tests

Table 24 summarises the mean treatment arm differences in the week 24 SPTs, adjusted for baseline value, age and total IgE level. Baseline and week 24 unadjusted SPT results are presented in Appendices 3 and 5.

TABLE 24 - Adjusted treatment arm difference in SPT reactivity (mm) at 24 weeks

Adjusted for baseline SPT, age and total IgE level.

p-value from a linear regression of week 24 SPT on treatment arm, adjusted for baseline SPT, age and total IgE level.

Test	Omalizumab arm (n)	Placebo arm (n)	Adjusteda mean difference in wheal size (mm): omalizumab – placebo (95% CI)	p-valueb
Cow’s milk (fresh)	19	16	–0.3 (–2.6 to 2.1)	0.820
Cow’s milk (extract)	17	16	–0.7 (–1.7 to 0.3)	0.187
Egg white	19	15	–5.6 (–9.4 to –1.9)	0.004
Egg white (raw)	18	13	–11.8 (–19.0 to –4.6)	0.002
Soya	19	15	–1.0 (–2.4 to 0.3)	0.137
Wheat	18	15	–1.2 (–2.5 to 0.1)	0.080
Peanut	19	15	–3.0 (–6.1 to 0.0)	0.050
Brazil nut	19	15	–2.0 (–4.2 to 0.1)	0.063
Hazelnut	19	15	–3.6 (–6.3 to –1.0)	0.010
Almond	19	15	–2.4 (–4.2 to –0.6)	0.012
Walnut	19	15	–3.0 (–5.4 to –0.5)	0.021
Cashew	19	15	–0.1 (–2.7 to 2.6)	0.960
Pistachio	19	15	–0.9 (–3.4 to 1.6)	0.464
Pecan	19	15	–1.4 (–3.0 to 0.2)	0.089
Macadamia	19	15	–1.1 (–2.9 to 0.8)	0.260
Sesame	18	15	–1.5 (–3.7 to 0.8)	0.198
Pine nut	18	15	–0.4 (–0.7 to 0.0)	0.049
Cod	19	15	–1.1 (–4.2 to 2.0)	0.487
Alternaria spp.	19	15	–2.0 (–4.1 to 0.0)	0.055
House dust mute (Dermatophagoides pteronyssinus)	22	16	–2.4 (–4.4 to –0.4)	0.023
House dust mute (Dermatophagoides farinae)	22	16	–2.7 (–5.1 to –0.2)	0.036
Birch pollen	20	17	–1.7 (–3.9 to 0.6)	0.147
Timothy grass pollen	20	17	–2.2 (–4.2 to –0.3)	0.027
Cat	18	15	–2.8 (–6.2 to 0.6)	0.107
Dog	18	15	–0.9 (–3.1 to 1.2)	0.382
Rabbit	18	15	0.2 (–0.6 to 1.1)	0.541
Horse	18	15	–0.2 (–1.5 to 1.1)	0.771
Shrimp	10	8	–1.2 (–3.0 to 0.7)	0.199

There were no strong imbalances in wheal sizes (mm) between the two arms at baseline, but they were generally smaller in the omalizumab arm than in the placebo arm at week 24. The number of positive SPTs (wheal > 3 mm) at week 24 is summarised by treatment arm in Figure 16.

FIGURE 16.

Distribution of the number of positive SPTs (wheal > 3 mm) at week 24 by arm. (a) Omalizumab; and (b) placebo.

The median number of positive SPTs (wheal > 3 mm) in the omalizumab arm at 24 weeks was 3.0 (IQR 1–7; minimum = 0 and maximum = 24; n = 23). In the placebo arm, the median number of positive SPTs was 10 (IQR 6–12; minimum = 1 and maximum = 19; n = 24). The unadjusted incidence rate of positive SPTs at week 24 was 4.91 (95% CI 4.09 to 5.91) in the omalizumab arm and 9.54 (95% CI 8.38 to 10.86) in the placebo arm. The unadjusted incidence rate ratio of positive SPTs for the omalizumab arm compared with the placebo arm was 0.51 (95% CI 0.41 to 0.65; p < 0.001), indicating a significant 49% decrease in the incidence rate of positive SPTs for omalizumab relative to placebo. A zero-inflated Poisson regression model was used to adjust the incidence rate ratio for the baseline number of positive SPTs, age and total IgE level. The adjusted incidence rate ratio for omalizumab (n = 22) compared with placebo (n = 17) was 0.56 (95% CI 0.40 to 0.78; p = 0.001), indicating a significant 44% decrease in the incidence rate of positive SPTs for omalizumab relative to placebo, given baseline number of SPTs, age and total IgE level.

Eczema exacerbations and infections

The overall numbers of eczema exacerbations and infections were low and there was no significant difference between arms.

Number of eczema exacerbations

Table 25 summarises the number of eczema exacerbations by treatment arm. The majority of participants in both treatment arms had no eczema exacerbations (83% omalizumab arm, 81% placebo arm). A total of five participants (17%) in the omalizumab arm and four participants (13%) in the placebo arm had one eczema exacerbation. No participants in the omalizumab arm had two or more exacerbations. A further two participants (6%) had two eczema exacerbations in the placebo arm (Figure 17). The incidence rate of eczema exacerbations over 24 weeks was 0.17 (95% CI 0.07 to 0.40) in the omalizumab arm and 0.26 (95% CI 0.13 to 0.52) in the placebo arm. The unadjusted incidence rate ratio of eczema exacerbations over 24 weeks for the omalizumab arm compared with the placebo arm was 0.65 (95% CI 0.21 to 1.97; p = 0.443), indicating an insignificant decrease in the incidence rate of eczema exacerbations over the treatment period for omalizumab relative to placebo. A zero-inflated Poisson regression model was used to adjust the incidence rate ratio for baseline age and total IgE level. The adjusted incidence rate ratio for omalizumab compared with placebo was 0.30 (95% CI 0.08 to 1.15; p = 0.080), indicating a non-significant decrease in the incidence rate of eczema exacerbations over the treatment period for omalizumab relative to placebo, given baseline age and total IgE level.

TABLE 25 - Number of eczema exacerbations

Note

The denominator includes participants followed up for the full 24-week follow-up period. It excludes one participant who withdrew from treatment and study participation following week 6 owing to transportation issues.

Number of eczema exacerbations	Treatment arm, n (%)
	Omalizumab (N = 30)	Placebo (N = 31)
0	25 (83)	25 (81)
1	5 (17)	4 (13)
2	0 (0)	2 (6)

FIGURE 17.

Number of eczema exacerbations. (a) Omalizumab; and (b) placebo.

Infective episodes of eczema

Table 26 summarises the number of infective eczema exacerbations by treatment arm. The majority of participants in both treatment arms had no infective eczema exacerbations (80% omalizumab arm, 74% placebo arm). One infective eczema exacerbation was experienced by five participants (17%) in the omalizumab arm and by six participants (19%) in the placebo arm. Two infective eczema episodes were experience by one participant (3%) in the omalizumab arm and by two participants (6%) in the placebo arm (Figure 18).

TABLE 26 - Number of infective eczema exacerbations

Note

The denominator includes participants followed up for the full 24-week follow-up period. It excludes one participant who withdrew from treatment and further study visits after week 6 owing to transportation issues.

Adapted from Chan et al. 81 Adapted with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Number of infective eczema exacerbations	Treatment arm, n (%)
	Omalizumab (N = 30)	Placebo (N = 31)
0	24 (80)	23 (74)
1	5 (17)	6 (19)
2	1 (3)	2 (6)

FIGURE 18.

Number of infective eczema exacerbations. (a) Omalizumab; and (b) placebo.

The incidence rate of infective episodes of eczema over 24 weeks was 0.23 (95% CI 0.11 to 0.49) in the omalizumab arm and 0.32 (95% CI 0.17 to 0.60) in the placebo arm. The unadjusted incidence rate ratio of infective episodes of eczema over 24 weeks for the omalizumab arm compared with the placebo arm was 0.72 (95% CI 0.28 to 1.90; p = 0.511), indicating a non-significant decrease in the incidence rate of infective episodes of eczema over the treatment period for omalizumab relative to placebo. A zero-inflated Poisson regression model was used to adjust the incidence rate ratio for baseline age and total IgE level. The adjusted incidence rate ratio for omalizumab compared with placebo was 0.53 (95% CI 0.12 to 2.24; p = 0.388), indicating a non-significant decrease in the incidence rate of infective episodes of eczema over the treatment period for omalizumab relative to placebo, given baseline age and total IgE level.

Serious adverse events and reactions

Table 27 summarises data by type of event for the safety set population, which includes all participants who received at least one injection of their allocated treatment (n = 62). Out of the 370 adverse events reported, a total of 15 were serious, of which one was a suspected serious adverse reaction.

The one serious adverse reaction occurred in a participant in the omalizumab arm (participant 1089). This participant had a history of recurrent idiopathic anaphylaxis and reactions preceding their enrolment in the study. The participant was well on discharge following their week 4 visit for their third treatment injection. Ten hours later, the participant developed an unexplained sudden onset of difficulty breathing, developed a cough, developed a wheeze, had difficulty swallowing, felt a lump in their throat and felt faint. They were administered an adrenaline autoinjector and their symptoms immediately started to improve. They were taken to hospital by ambulance and were discharged after a 6-hour period of observation. Following a discussion with the DMEC, during which all parties remained blinded to the treatment allocation, participant 1089 was withdrawn from the study as it was not possible to exclude a reaction to the IMP or placebo. The participant went on to have two further unexplained episodes of anaphylaxis, 2 days apart at week 25, after they had been off the study treatment for 21 weeks. The participant’s treatment allocation remained blinded until the full data set was unblinded for all patients.

In both treatment arms, a total of seven SAEs occurred in six participants. Details of all the serious events are given in Table 28. One of the criteria for defining an adverse event as serious was the requirement for hospitalisation during the episode. There were more participants who experienced eczema exacerbations requiring hospitalisation in the placebo arm (three participants) than in the active omalizumab arm (one participant).

Non-serious adverse events

A full listing of the non-serious adverse events and reactions by MedDRA-preferred term and treatment arm is given in Appendix 6, and Table 29 summarises the non-serious adverse events by body system class.

There was a notably lower number of participants experiencing respiratory and dermatological events in the omalizumab arm than in the placebo arm. A total of 15 participants (50%) experienced respiratory events in the omalizumab arm, compared with 25 (78%) in the placebo arm (relative risk 0.64, 95% CI 0.43 to 0.96); 23 participants (77%) experienced dermatological events in the omalizumab arm, compared with 31 (97%) in the placebo arm (relative risk 0.79, 95% CI 0.64 to 0.98).

Figure 19 is a volcano plot of the non-serious adverse events by MedDRA-preferred term. For each event, the risk difference between the treatment arms is plotted against the p-value from a Fisher’s exact test. The exact significance of the test results should be interpreted with caution because these results do not take into account multiple testing and will be underpowered to detect meaningful differences, but the p-value does provide a measure of the strength of the evidence for a difference. An examination of the events with the largest risk difference and lowest p-value by treatment arm at MedDRA-preferred term highlights a lower risk of exacerbations of asthma, infective eczema, upper respiratory tract infection and coughs in the omalizumab arm and it highlights a lower risk of headaches, iron deficiency and allergic reaction in the placebo arm.

FIGURE 19.

Volcano plot of the non-serious adverse events by MedDRA-preferred term. URTI, upper respiratory tract infection. Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Reproduced from Chan et al. 81 Reproduced with permission from JAMA Pediatrics. 2020. 174(1): 29–37. Copyright © (2020) American Medical Association. All rights reserved.

Eczema severity

The mean treatment arm difference in objective SCORAD score at 24 weeks (the primary outcome) was significant, in favour of omalizumab, at –6.9 (95% CI –12.2 to –1.5; p = 0.013). Although the point estimate was lower than the MCID, the CI includes the previously reported MCID of 8.2 as well as the MCID of 8.5 that was calculated for the specific population seen in ADAPT. The within-arm treatment difference exceeded the MCID for the omalizumab arm only (–12.4). The mean treatment arm differences over time (see Figure 3) demonstrate a widening of the difference in outcome between the active and placebo arms up to 24 weeks, at which point the treatment was discontinued. This suggests that the treatment effect did increase over time and may have persisted if treatment was continued past the 24-week limit of the trial. This is in line with the findings and suggestions emerging from other studies that longer treatment duration may accrue greater benefits. 51,56 Further exploration in sensitivity analyses that adjusted for alternative therapies, explored the impact of missing data and adherence, found that these factors made little difference to the outcome at 24 weeks, adding weight to the robustness of the results. Beyond the week 24 treatment period, the difference between the arms in outcome reduced. At week 48, the treatment difference remained, on average, in favour of the omalizumab arm (–2.8, 95% CI –8.6 to 3.0; p = 0.346); however, the 95% CI includes the possibility that there was no longer a difference or a favouring of placebo at 48 weeks (see Figure 3).

The other eczema severity assessments, the total SCORAD and the EASI, both demonstrated significant treatment arm differences in agreement with the objective SCORAD. The total SCORAD average treatment effect was close to the MCID; for EASI, the estimate surpassed the MCID. The within-arm assessments exceeded the MCID for only the omalizumab arm. In the total SCORAD and EASI assessments, the gap between the omalizumab and placebo arms remains, although to a lesser extent, towards week 48 (see Figures 7 and 8).

These results suggest an improvement in eczema severity in participants treated with omalizumab compared with those treated with placebo. The effect was significant given that the median baseline IgE level in ADAPT participants was 8373 kU/l, which is 120 times the upper limit of the normal range of 70 kU/l for total IgE level and 5.6 times higher than the maximum dose of omalizumab is designed to treat.

In contrast, the treatment effect observed on the patient-derived eczema severity measure, the POEM, was smaller and did not achieve the reported MCID of 3.4. Anecdotally, it was noted that children who complete the POEM themselves would score lower than if the scoring was completed by their parents. It was speculated that this could be a result of the child’s acceptance of the status quo in a chronic condition, without the experience of life without eczema. Alternatively, a parent, who may not have experienced eczema themselves, may view the experience from a different perspective, either by being more balanced or more aware of the significant difficulties that their child accepts as part of the course. There are no official age cut-off points for proxy completion of the POEM,74 but it is recognised that older children and young people would often complete the scoring themselves, whereas parents would be involved in the scoring of younger children.

We therefore used a cut-off age of 10 years (the approximate average age of participants in the study), assuming that children and young people over 10 years of age would complete the questionnaires themselves, to assess if there was a genuine difference in scores of older and younger children. This analysis indeed showed a larger and significant treatment effect in younger children, with an average treatment effect of –5.2 (95% CI –10.0 to –0.5; p = 0.031) in favour of omalizumab. In contrast, the treatment effect in older children and young people indicated an average increase in score (or worsening of QoL) for those in the omalizumab arm in comparison with those in the placebo arm, and this was non-significant, at 2.8 (95% CI –1.8 to 7.4; p = 0.230).

It could be that younger children may genuinely benefit more than older children and young people from treatment with omalizumab; however, there was no difference in the treatment effect as assessed by the objective or total SCORAD between the two different age groups (data not shown). An alternative explanation is that the older children and young people who completed their own questionnaires underestimated any treatment effect, and this may explain why POEM scores were at odds with the other results.

Treatment burden and use of other medication: potent topical steroid and systemic therapy

Participants in the study were allowed to continue using their topical therapies, including potent topical steroids, ad libitum. This was a cohort of participants with severe eczema, who had access to little therapy that provided them with relief from their symptoms. The concomitant use of potent topical steroids reflects practice in the real world. Furthermore, withdrawing routine treatment for up to 1 year, particularly as half of the participants were taking a placebo medication, would have been challenging for their management and for participant retention.

Interestingly, the median number of days of potent topical steroid use was 48% higher at week 24 and 55% higher at week 48 in the placebo arm than in the omalizumab arm. The median BSA coverage was also 2.0 times higher for participants in the placebo arm than for the omalizumab-treated participants at 24 weeks, and 1.7 times higher for participants in the placebo arm at 48 weeks in comparison with the omalizumab arm. The median total weight of potent topical steroid use was 76% higher in the placebo arm at both 24 and 48 weeks. Over the 24-week treatment period, the participants in the placebo arm were using 76% more potent topical steroids on most days [median of 161 days out of the total of 168 days (24 weeks) of the study] on a median of one-third of their BSA (see Table 23). This is a very significant burden of potent topical steroid use. After active treatment was discontinued at 24 weeks, the difference remained when assessed at week 48, although the gap had started to close on BSA potent topical steroid use between the arms. This was despite a similar level of potent topical steroid use at baseline in the two arms (see Table 3).

Topical therapy usage is notoriously difficult to assess in clinical trials but we felt that it was important to attempt to record this. Assessing the weight of tubes of creams used was as challenging as it has been in previous studies, especially as families had multiple tubes in different locations and did not have their medication at every hospital visit. Thus, in addition to this and as a proxy measure, we asked families to tell us on how many days they had used potent topical steroids creams since their last visit, and over what areas of their body. Although potentially subject to parental or participant recall, this method of calculating potent topical steroid usage was easily adopted by clinicians using the same method to assess BSA for the SCORAD and gives a good representation of topical steroid utilisation. When compared with the more limited data that we had on the weight of potent topical steroid used, both sets of data concurred that there was much more potent topical steroid used in the placebo arm. Although we were able to collect the data for many more participants using days used and BSA covered, this was still a limited dataset, and it should be interpreted in this context.

When we consider the treatment burden as treatment failure, defined as requiring at least two courses of rescue therapy (with oral prednisolone), or requiring AST (systemic immunosuppression by week 30), there was a total of one participant in the omalizumab arm (3.3%) and five in the placebo arm (16.1%) who met these criteria. Three participants in the placebo arm were defined as treatment failures requiring at least two courses of rescue therapy with oral prednisolone, and one participant from the omalizumab arm fell into this category. Four participants in the placebo arm received alternative systemic immunosuppressive eczema therapy (AST) by week 30. The only participant from the omalizumab arm who received AST (methotrexate) had been withdrawn from receiving further study treatment by the study investigators by week 6. By week 48, six participants in each arm had initiated treatment.

Taken together with the improvement in eczema severity scores, it appears that the eczema severity of participants in the omalizumab arm improved despite lower potent topical steroid use than their placebo counterparts. A combined symptom–medication score would have been a very useful adjunct in this study, although we are not aware of any validated measures in clinical use. In addition, there is also a suggestion of a lower systemic treatment load in participants, with fewer courses of oral steroids and less systemic therapy prescribed for eczema.

The reduced drug burden is a potentially significant finding as potent topical steroids and systemic immunosuppression are not without side effects. Families are often concerned about the use of topical and oral steroids and the current arsenal of systemic eczema treatments for children owing to their implications for the short- and long-term health of their children. The burden of therapy can, in some cases, be as disabling as the burden of disease. Although the numbers are small and not statistically significant, this is a clinically important finding that, if borne out in further studies, would be of significance to patients with severe eczema and their families.

Quality of life

Two QoL assessments were employed in this study to address different aspects of the burden of disease. The (C)DLQI assesses QoL in dermatological disease in general and the PADQLQ assesses the full systemic picture of allergic disease.

The (C)DLQI achieved a –3.5-point difference, which was significant and clinically relevant, surpassing the reported MCID of 3.3 for the DLQI. The mean (C)DLQI score halved from 17 (SD 5.6) at baseline to 8.5 (SD 5.9) at week 24 in the omalizumab arm. This is an important and striking change. In adult patients, at least a 5-point reduction in DLQI is one of the NICE criteria86 used to decide if systemic therapy should be continued in psoriasis, which this achieves.

There was also a clinically significant effect on the PADQLQ, with marginal statistical significance.

Assessment of allergic markers

There was a significant 50% reduction in the number of positive SPTs at 24 weeks in the omalizumab arm compared with the placebo arm. This result did not appear to be reflected in the SpIgE results.

The finding of a reduction in SPT with omalizumab treatments has been reported previously. 87 Although SPT results can vary over time without intervention, improvement in SPT reactivity over such a short period (of 24 weeks) may suggest a biological effect, despite the very raised total IgE levels in this patient population. It may herald an effect on other diseases, such as food allergy and rhinoconjunctivitis, which was beyond the scope of this study to fully assess. Taken in conjunction with the improvement in the PADQLQ, a system-wide QoL assessment of allergic disease in children, this could suggest a potential role for omalizumab in multisystem allergic disease.

Adverse events and adverse reactions

Of the non-serious adverse events, participants on active treatment appeared to have fewer respiratory and dermatological events (see Table 29). This was despite a balanced profile of pre-existing conditions, including respiratory and dermatological conditions recorded at baseline (see Appendix 3). The association with fewer respiratory events may be another pointer to the multisystem effects of omalizumab. It is already well established in its role in asthma and chronic urticaria, and studies have shown its efficacy in rhinoconjunctivitis. The improvements in PADQLQ, which reflects the multiple systems affected by allergic disease, also support this hypothesis.

The overall numbers of eczema exacerbations and infections were low and although there was a marginal increase in these events in the placebo arm, this did not reach statistical significance. Although the overall numbers of SAEs were matched in both arms, three of the participants in the placebo arm required admission to hospital to manage their exacerbations, compared with one participant in the active omalizumab arm. Hospital admission qualifies the episode as a SAE. This may be a potentially important indication of the degree of eczema control afforded by treatment with omalizumab, although numbers remain too small for conclusions to be drawn from this observation.

There was one potential severe adverse reaction reported in participant 1089. The reaction took place 10 hours following their discharge from hospital for their third treatment injection. All parties remained blinded to treatment allocation when the participant was withdrawn from the study by investigators. It transpired at unblinding of the entire dataset that the participant was in receipt of the active omalizumab treatment.

The participant had a background history of recurrent idiopathic anaphylaxis and multiple weekly allergic reactions. They reported two further episodes of unexplained anaphylaxis within the study period, 21 weeks after treatment had been discontinued, suggesting that these later events were unrelated to omalizumab, but making it difficult to be certain of the aetiology of the index episode. The largest cohort of omalizumab-related anaphylaxis, with 132 cases, was published in December 2017. 88 This suggests that anaphylaxis often occurs within the first three doses (72%), in patients who have a history of anaphylaxis in response to other triggers, in female patients (84%) and with more than half of the cases occurring within an hour of drug administration. Participant 1089 fulfilled many of these criteria but reported a delayed time to onset of symptoms; however, delayed anaphylaxis is not unknown in omalizumab treatment. Thus, the possibility that this was an adverse drug reaction and not an episode in line with the participant’s previous history of idiopathic anaphylaxis cannot be excluded. This supports the need for caution when prescribing omalizumab in patients who meet these high-risk criteria.

These results suggest that there may be lower rates of eczema-related and respiratory non-serious adverse events with omalizumab. The number of severe adverse events was the same between the arms. One of the participants reported an episode of anaphylaxis, which was potentially an adverse reaction to omalizumab.

Clinical study team

Muhsinah Adam, Freda Adjei, Dr Susan Chan, Louise Coverdale, Helen Fisher, Joanna Gambell, Claire Hearn, Jemma Jackson, Rahi Jahan, Christine Jameson, Patricia Kane, Colm Keenan, Gary McGreevy, Una O’Dwyer-Leeson, Victoria Offord, Dr Suzana Radulovic, Anjum Shah, Prabalini Thaventhiran, Victoria Timms, Fiona Watson and Dr Emma Wedgeworth.

Trial Steering Committee

We thank Dr Nerys Roberts (Chairperson), Dr Mike Coren, Dr Helen Cox, Professor John Harper, Mrs Kathryn Humphreys, Dr Emma Wedgeworth, Professor Gideon Lack, Dr Victoria Cornelius, Dr Susan Chan, for their advice and contributions to the protocol and statistical analysis plan. We also wish to thank Suzie Cro, Tao Chen and Kun Liu for their important contribution to the development of the statistical methods.

Data Monitoring and Ethics Committee

Professor Graham Roberts (Chairperson), Dr Ahmed Massoud and Helen Moyses.

Statistics team

Dr Victoria Cornelius, Dr Suzie Cro, Dr Tao Chen and Dr Kun Liu.

Laboratory team

Rosemary Ebling, Ewa Pietraszewicz, Dr Alick Stephens, Asha Sudra and Dr Victor Turcanu.

Medical photography

Nigel Pearson and Gary Mulcahy.

Participant identification centres

Barts Health NHS Trust (Dr Sasha Dhoat, Professor Jonathan Grigg, Tori Hadaway and Dr David Paige), Chelsea and Westminster Hospital NHS Foundation Trust (Dr Peggy Chen, Dr Charlotte Edwards, Dr Bisola Laguda, Dr Eirini Merika, Dr Eugene Ong and Dr Nerys Roberts), East Kent Hospitals University NHS Foundation Trust (Dr Susannah Baron), Great Ormond Street Hospital for Children NHS Foundation Trust (Dr Despina Eleftheriou, Dr Georgios Eleftheriou, Dr Mary Glover, Dr Karolina Gholam, Debra Lomas, Dr Caroline Mahon and Dr Anna Martinez), Guy’s and St Thomas’ NHS Foundation Trust (Dr Aaron Boyce, Dr Helen Brough, Dr Natalia Cartledge, Professor George Du Toit, Roisin Fitzsimons, Dr Sophie Flammarion, Dr Carsten Flohr, Dr Adam Fox, Dr Eleanor Higgins, Professor Gideon Lack, Dr Eleanor Mallan, Dr Tom Marrs, Dr Jemima Mellerio, Emily Rodger, Dr Alexandra Santos, Heather Sharp, Professor Catherine Smith, Dr Kate Swan, Dr Lauri-Ann Van der Poel, Dr Rosy Wells, Dr Alya Abdul-Wahab, Dr Emma Wedgeworth and Dr Stephanie Weston), Imperial College Healthcare NHS Trust (Dr Catherine Hardman), King’s College Hospital NHS Foundation Trust (Dr Marie-Louise Daly, Dr Sandy Flann, Dr Elisabeth Higgins, Dr Rachael Morris-Jones, Dr Theng Theng Lew and Dr Laura Proudfoot), Lewisham and Greenwich NHS Trust (Dr Shamali Hoque and Dr Thomas Tull), Royal Free London NHS Foundation Trust (Dr Danielle Greenblatt), St George’s University Hospitals NHS Foundation Trust (Dr Montserrat Gilaberte Pena, Dr Michael Perkin and Dr Sara Sherif), West Hertfordshire Hospitals NHS Trust (Dr Kapila Batta) and The Whittington Hospital NHS Trust.

Tertiary, secondary and private referrals

Brighton and Sussex University Hospitals NHS Trust (Dr Jessie Felton), Epsom and St Helier University Hospitals NHS Trust (Dr Natalia Spierings), Hampshire Hospitals NHS Foundation Trust (Dr Satish Chandran, Dr Priya Ilangovan and Dr Turner), Heart of England NHS Foundation Trust (Dr Helen Goodyear), Kingston Hospital NHS Foundation Trust (Dr Claire Fletcher), Luton and Dunstable University Hospital NHS Foundation Trust (Dr Beryl Adler and Dr Bernadette De Silva), Maidstone and Tunbridge Wells NHS Trust (Dr Uday Kumar), Oxford University Hospitals NHS Foundation Trust (Dr Richard Turner), Royal Surrey County Hospital NHS Foundation Trust, University Hospital Southampton NHS Foundation Trust and private referrals (Dr David Atherton, Professor George Du Toit, Dr Adam Fox, Dr Mary Glover, Professor John Harper, Professor Gideon Lack and Dr Malcolm Rustin).

King’s College London and Guy’s and St Thomas’ NHS Foundation Trust

Elizabeth Bruna, Helen Critchley, Colin Hutchinson, Karen Ignatian, Salma Kibaida, Maria Mason, Hinna Mir, Samantha Roper, Emma Tylor, Mandy Wan and Andrew Webb.

King’s Clinical Trials Unit

Joanna Kelly and Caroline Murphy.

Others

Oliver Gupta (Modepharma).