Intermittent montelukast in children aged 10 months to 5 years with wheeze (WAIT trial): a multicentre, randomised, placebo-controlled trial

Summary Background The effectiveness of intermittent montelukast for wheeze in young children is unclear. We aimed to assess whether intermittent montelukast is better than placebo for treatment of wheeze in this age group. Because copy numbers of the Sp1-binding motif in the arachidonate 5-lipoxygenase (ALOX5) gene promoter (either 5/5, 5/x, or x/x, where x does not equal 5) modifies response to montelukast in adults, we stratified by this genotype. Methods We did this multicentre, parallel-group, randomised, placebo-controlled trial between Oct 1, 2010, and Dec 20, 2013, at 21 primary care sites and 41 secondary care sites in England and Scotland. Children aged 10 months to 5 years with two or more wheeze episodes were allocated to either a 5/5 or 5/x+x/x ALOX5 promoter genotype stratum, then randomly assigned (1:1) via a permuted block schedule (size ten), to receive intermittent montelukast or placebo given by parents at each wheeze episode over a 12 month period. Clinical investigators and parents were masked to treatment group and genotype strata. The primary outcome was number of unscheduled medical attendances for wheezing episodes. Analysis was by intention to treat. This trial is registered with ClinicalTrials.gov, number NCT01142505. Findings We randomly assigned 1358 children to receive montelukast (n=669) or placebo (n=677). Consent was withdrawn for 12 (1%) children. Primary outcome data were available for 1308 (96%) children. There was no difference in unscheduled medical attendances for wheezing episodes between children in the montelukast and placebo groups (mean 2·0 [SD 2·6] vs 2·3 [2·7]; incidence rate ratio [IRR] 0·88, 95% CI: 0·77–1·01; p=0·06). Compared with placebo, unscheduled medical attendances for wheezing episodes were reduced in children given montelukast in the 5/5 stratum (2·0 [2·7] vs 2·4 [3·0]; IRR 0·80, 95% CI 0·68–0·95; p=0·01), but not in those in the 5/x+x/x stratum (2·0 [2·5] vs 2·0 [2·3]; 1·03, 0·83–1·29; p=0·79, pinteraction=0·08). We recorded one serious adverse event, which was a skin reaction in a child allocated to placebo. Interpretation Our findings show no clear benefit of intermittent montelukast in young children with wheeze. However, the 5/5 ALOX5 promoter genotype might identify a montelukast-responsive subgroup. Funding Medical Research Council (UK) and National Institute for Health Research.

polymorphism in the promoter of ALOX5 (rs59439148) was genotyped as previously analysed by capillary electrophoresis on a 3130xl Genetic Analyser (Applied Biosystems). Fragments of 256 to 292bp were obtained depending on the copy number (2 to 8) of the repeat sequence and were visualized using GeneMapper v4.0 or Peak Scanner v1.0 software. Genotypes were called manually from duplicate amplifications. Samples with known genotypes, which had been verified by DNA sequence analysis, were included in each run. Alleles were called according to the number of simple repeats. Samples with the most common genotype (homozygous 5/5) were allocated to one stratum. Samples with any other genotype (either 5/x or x/x, where x is any allele other than 5 were allocated to the [5/x+x/x] stratum.

Urinary leukotriene E 4
Urine samples were obtained from children immediately following sputum induction and stored on ice until transfer to storage at -80 o C within 1 h. Analysis was by gas chromatography-mass spectrometry using negative ion chemical ionisation (GC-NICI-MS). To each urine sample (0.5 mL), chemically identical deuterated internal standards (1ng, Cayman Chemical Company, AnnArbor, MI) were added, acidified with acetic acid (pH 4.5), then extracted twice with methyl tertbutyl-ether and dried under nitrogen. Methanol dissolved aliquots (10µL, in methanol) were injected onto a reverse phase column (Zorbax Eclipse XDB C-18, Agilent Technologies, Inc. Santa Clara, CA, USA), stabilised thermally at 37 o C and a gradient consisting of two mobile phases: A acetonitrile/water/acetic acid (20/80/0.0001) and B acetonitrile/isopropanol/acetic acid (55/45/0.0001, v/v) was used to elute LTE4 with the flow rate 0.11 mL/min using HPLC equipped with an autosampler (Shimadzu Sil-2-AC, Shimadzu Scientific Instruments, Inc. Columbia, MD, USA). The mobile phase binary linear gradient was 1 min 8% B, 9.5 min 8 to 95% B, 0.5 min 95% B, 0.5 min 95-100% B, 2 min 100% B. Leukotriene E 4 was measured using multiple reaction monitoring mode (MRM) tandem mass spectrometry (Qtrap 4000, Applied Biosystems, Foster City, CA, USA) equipped with an electrospray ion source. Urine sample extract was prepared by a 2 step derivatisation to pentafluorobenzyl and trimethylsilyl esters which modified carboxyl and hydroxyl groups of the compound, and were purified by a thin-layer chromatography. A gas chromatography negativeion chemical ionisation mass-spectrometry was used for quantification (model Engine 5989B series II Helwett Packard, Palo Alto, CA). All solvents were HPLC grade and purchased from Mallincrodt Baker, Inc. Phillipsburg, NJ, USA), while other chemicals were from Sigma-Aldrich Co. St. Louis, MO, USA).

Combined analysis
In order to compare the primary outcome of this study to published reports, we undertook a review of previously published data based on the non-systematic review approach described by Ducharme and colleagues 2 for the period January 1980 to January 2014. Two authors (JG and RB) searched EMBASE, SCOPUS, MEDLINE and the Cochrane Airways Group trials register for details of trials published after the review by Ducharme and colleagues 2 . We sought further trials from 1 st January 2014 to 1 st August 2014 with a placebo-controlled design that assessed the efficacy of intermittent montelukast with respect to unscheduled medical attendances for wheezing in pre-school children over a 12-month follow-up. We identified no additional studies were identified. The three studies reported by Ducharme and colleagues 2 were therefore selected for the combined analysis; Robertson and colleagues 3 , Valovirta and colleagues 4 , and Bacharier and colleagues 5 .
The published reports were reviewed for details on pre-specified outcomes. The study by Robertson and colleagues 3 reported "unscheduled health care resource utilisations (HRU) in the group of children who received at least one episode of treatment as a primary outcome", in children between two to 14 years of age. Raw data from this study were provided by personal communication (by author DP), from which we calculated the mean and standard deviation (SD) for the number of episodes requiring an unscheduled medical attendance for each child within the subgroup of children aged two to five years. The study by Bacharier and colleagues 5 enrolled children aged one to six years, with number of "unscheduled visits to primary care offices, urgent care or emergency departments and hospitalisations for acute wheezing episodes" as a secondary outcome. The mean and standard deviation for these episodes was published in their report, additionally the number of children with one or more events for this outcome was confirmed by personal correspondence (Dr Bacharier). Original data were not received from the study of Valovirta and colleagues 4 which included children aged six months to five years, and reported a secondary outcome for "adjusted rate of asthma attacks", with an attack defined within the statement; "the start of an asthma attack was the first day the patient's symptoms required HRU". Overall, the 3 studies were comparable in terms of study design (placebo-controlled trial), randomisation, concealment of allocation, inclusion criteria and duration of follow-up (12 months) to those in this report (Supplementary Table 7).
A meta-analysis was performed using the inverse variance fixed effect method to calculate the summary weighted difference between the means with 95% confidence intervals. The analysis was performed using RevMan version 5.3 6 . We compared the mean (SD) for number of episodes requiring an unscheduled medical attention per child in each study, comparing experimental (intermittent montelukast) and control (placebo) groups, using a fixed effects model for mean difference as per the review by Ducharme and colleagues 2 . The overall analysis of 2783 children shows no overall benefit for intermittent montelukast therapy in reducing need for unscheduled medical attention for wheeze (weighted mean difference = -0.10, 95% CI -0.26 to -0.06, test for overall effect p=0.21, Fig. S2, Appendix).

Please answer as well as you can remember
Please answer the questions by ticking ( ) Yes or No Y N Did your child wheeze in the last 24 hours? Did your child have a cold in the last 24 hours? Did you give your child the TRIAL medicine TODAY? Did your child vomit the medicine TODAY? Did your child miss school or nursery TODAY? Did ANYONE stay home to look after your child TODAY? Did your child see a doctor or nurse TODAY? Did you give your child the blue inhaler in the last 24 hrs If yes? How many times did you give it to them in the last 24 hours? On average, how many puffs did you give them each time?       OR; odds ratio, IRR incidence rate ratio. wheeze with and without viral colds, Episodic wheeze; wheeze only with colds Table S8. Relationship of adverse events to study medication.  ED; emergency department, HRU; health care resource utilisation, RCT; randomised controlled trial, RTI; respiratory tract infection Figure S1. Study design and schedule of procedures.
Primary outcome data were taken from the phonecall (T), which occurred every two months. ALOX5; arachidonate 5-lipoxygenase gene promoter, IMP; investigational medicinal product, USMA; need for unscheduled medial attention for wheeze, V; visits to study team The pooled summary (diamond) represents the overall result, with the vertical line representing no overall effect (weighted mean difference=0). There is no benefit of intermittent montelukast in an unselected population of children with preschool wheeze (test for overall effect, p=0·21). Nwokoro 2014 refers to the current study