BACKGROUND: Despite inhaled corticosteroid plus long-acting β-agonist (ICS/LABA) therapy, 30-50% of patients with moderate or severe asthma remain inadequately controlled. We investigated the safety and efficacy of single-inhaler fluticasone furoate plus umeclidinium plus vilanterol (FF/UMEC/VI) compared with FF/VI. METHODS: In this double-blind, randomised, parallel-group, phase 3A study (Clinical Study in Asthma Patients Receiving Triple Therapy in a Single Inhaler [CAPTAIN]), participants were recruited from 416 hospitals and primary care centres across 15 countries. Participants were eligible if they were aged 18 years or older, with inadequately controlled asthma (Asthma Control Questionnaire [ACQ]-6 score of ≥1·5) despite ICS/LABA, a documented health-care contact or a documented temporary change in asthma therapy for treatment of acute asthma symptoms in the year before screening, pre-bronchodilator FEV between 30% and less than 85% of predicted normal value, and reversibility (defined as an increase in FEV of ≥12% and ≥200 mL in the 20-60 min after four inhalations of albuterol or salbutamol) at screening. Participants were randomly assigned (1:1:1:1:1:1), via central based randomisation stratified by pre-study ICS dose at study entry, to once-daily FF/VI (100/25 μg or 200/25 μg) or FF/UMEC/VI (100/31·25/25 μg, 100/62·5/25 μg, 200/31·25/25 μg, or 200/62·5/25 μg) administered via Ellipta dry powder inhaler (Glaxo Operations UK, Hertfordshire, UK). Patients, investigators, and the funder were masked to treatment allocation. Endpoints assessed in the intention-to-treat population were change from baseline in clinic trough FEV at week 24 (primary) and annualised moderate and/or severe asthma exacerbation rate (key secondary). Other secondary endpoints were change from baseline in clinic FEV at 3 h post-dose, St George's Respiratory Questionnaire (SGRQ) total score, and ACQ-7 total score, all at week 24. Change from baseline in Evaluating Respiratory Symptoms in Asthma total score at weeks 21-24 was also a secondary endpoint but is not reported here. Exploratory analyses of biomarkers of type 2 airway inflammation on treatment response were also done. This study is registered with ClinicalTrials.gov, NCT02924688, and is now complete. FINDINGS: Between Dec 16, 2016, and Aug 31, 2018, 5185 patients were screened and 2439 were recruited and randomly assigned to FF/VI (100/25 μg n=407; 200/25 μg n=406) or FF/UMEC/VI (100/31·25/25 μg n=405; 100/62·5/25 μg n=406; 200/31·25/25 μg n=404; 200/62·5/25 μg n=408), with three patients randomly assigned in error and not included in analyses. In the intention-to-treat population, 922 (38%) patients were men, the mean age was 53·2 years (SD 13·1) and body-mass index was 29·4 (6·6). Baseline demographics were generally similar across all treatment groups. The least squares mean improvement in FEV change from baseline for FF/UMEC/VI 100/62·5/25 μg versus FF/VI 100/25 μg was 110 mL (95% CI 66-153; p<0·0001) and for 200/62·5/25 μg versus 200/25 μg was 92 mL (49-135; p<0·0001). Adding UMEC 31·25 μg to FF/VI produced similar improvements (FF/UMEC/VI 100/31·25/25 μg vs FF/VI 100/25 μg: 96 mL [52-139; p<0·0001]; and 200/31·25/25 μg vs 200/25 μg: 82 mL [39-125; p=0·0002]). These results were supported by the analysis of clinic FEV at 3 h post-dose. Non-significant reductions in moderate and/or severe exacerbation rates were observed for FF/UMEC 62·5 μg/VI versus FF/VI (pooled analysis), with rates lower in FF 200 μg-containing versus FF 100 μg-containing treatment groups. All pooled treatment groups demonstrated mean improvements (decreases) in SGRQ total score at week 24 compared with baseline in excess of the minimal clinically important difference of 4 points; however, there were no differences between treatment groups. For mean change from baseline to week 24 in asthma control questionnaire-7 score, improvements (decreases) exceeding the minimal clinically important difference of 0·5 points were observed in all pooled treatment groups. Adding UMEC to FF/VI resulted in small, dose-related improvements compared with FF/VI (pooled analysis: FF/UMEC 31·25 μg/VI versus FF/VI, -0·06 (95% CI -0·12 to 0·01; p=0·094) FF/UMEC 62·5 μg/VI versus FF/VI, -0·09 (-0·16 to -0·02, p=0·0084). By contrast with adding UMEC, the effects of higher dose FF on clinic trough FEV and annualised moderate and/or severe exacerbation rate were increased in patients with higher baseline blood eosinophil count and exhaled nitric oxide. Occurrence of adverse events was similar across treatment groups (patients with at least one event ranged from 210 [52%] to 258 [63%]), with the most commonly reported adverse events being nasopharyngitis (51 [13%]-63 [15%]), headache (19 [5%]-36 [9%]), and upper respiratory tract infection (13 [3%]-24 [6%]). The incidence of serious adverse events was similar across all groups (range 18 [4%]-25 [6%)). Three deaths occurred, of which one was considered to be related to study drug (pulmonary embolism in a patient in the FF/UMEC/VI 100/31·25/25 μg group). INTERPRETATION: In patients with uncontrolled moderate or severe asthma on ICS/LABA, adding UMEC improved lung function but did not lead to a significant reduction in moderate and/or severe exacerbations. For such patients, single-inhaler FF/UMEC/VI is an effective treatment option with a favourable risk-benefit profile. Higher dose FF primarily reduced the rate of exacerbations, particularly in patients with raised biomarkers of type 2 airway inflammation. Further confirmatory studies into the differentiating effect of type 2 inflammatory biomarkers on treatment outcomes in asthma are required to build on these exploratory findings and further guide clinical practice. FUNDING: GSK.