• Hector Ortega, Hao Li, Robert Suruki, Frank Albers, David Gordon, and Steven Yancey.
• 1 Respiratory Therapy Area Unit, Medicines Development Center, and.
• Ann Am Thorac Soc. 2014 Sep 1;11(7):1011-7.

RationaleDetailed characterization of asthma phenotypes is essential for identification of responder populations to allow directed personalized medical intervention.ObjectivesThe aim of this study was to identify distinctive patient characteristics within subgroups of a well-characterized severe asthma population at risk for exacerbations and to determine the treatment response within each subgroup.MethodsA supervised cluster analysis with recursive partitioning approach was applied to data from the Dose Ranging Efficacy And safety with Mepolizumab (DREAM) study to identify characteristics that maximized the differences across subgroups. Exacerbation rate ratios were calculated for each cluster comparing mepolizumab versus placebo.Measurements And Main ResultsThree predictors were identified in four primary clusters: blood eosinophils, airway reversibility, and body mass index. The reduction in exacerbations was significantly greater in patients who received mepolizumab (clusters 2, 3, and 4) with raised eosinophils (responder population). Cluster 2 with low airway reversibility (mean, 11%) had a 53% reduction in exacerbations. These patients more frequently reported sinusitis and nasal polyposis. Those with higher airway reversibility (mean, 28%) were further split by body mass index. The nonobese versus obese (clusters 3 and 4) had a 35 and 67% reduction in exacerbations, respectively. Cluster 4 also had patients with more comorbidities, including hypertension, weight gain, and anxiety.ConclusionsUsing supervised cluster analysis helped identify specific patient characteristics related to disease and therapeutic response. Patients with eosinophilic inflammation received significant therapeutic benefit with mepolizumab, and responses differed within clusters. Clinical trial registered with www.clinicaltrials.gov (NCT01000506).

36 keywords...

hide…