Cluster analysis of severe life-threatening asthma: identification of a distinct phenotype characterized by high serum eosinophils

Background: Severe life-threatening asthma (SLTA) has heterogeneous clinical phenotypes Aim: To identify clinically distinct SLTA phenotypes and compare their outcomes Methods: A retrospective study of patients admitted to Singapore General Hospital intensive care unit in 2010-2015 for “asthma exacerbation” or “status asthmaticus” and requiring intubation. Principal components analysis followed by two-step cluster analysis were used to identify distinct SLTA phenotypes. Time to next exacerbation requiring admission was analysed with Cox regression. Results: 55 cases were included. Median age (interquartile range, IQR) was 60 (45-72), 23 (41.8%) were male. 33 (60%) were on Global Initiative of Asthma step 4 or 5 treatment. Median follow-up was 1035d (IQR 526-1622d). Two distinct clusters were identified. Cluster 1 (n=17, 30.9%) had a higher absolute eosinophil count (median 0.79x10 9 /L, IQR 0.57-1.44x10 9 /L). Cluster 2 (n=38, 69.1%) had a lower absolute eosinophil count (median 0.24x10 9 /L, IQR 0.01-0.77x10 9 /L). Cluster 1 patients were younger (46 vs 63 yrs, p=0.029), more likely to be male (70.6% vs 28.9%, p=0.004), have greater smoking history (20 vs 0 pack-yrs, p=0.005), lower pH (7.00 vs 7.25, p Conclusion: We have identified two distinct SLTA phenotypes. A phenotype characterized by high serum eosinophils, more severe acidaemia and ventilatory failure, and an increased risk of future admissions.