Abstract
Background: Severe asthma is a heterogeneous condition that requires deeper clinical and biological phenotyping. This can be addressed by combining clinical and several omics data in a phenotypic handprint.
Objective: Sputum data from cell transcriptomics, somalogic proteomics and eicosanoids lipidomics from 73 adult U-BIOPRED patients, have been integrated, using a systems biology approach. Patient clusters supported by multiple data types were generated, along with specific biomarkers, defining a sputum handprint.
Methods: The three omics datasets were fused using the Similarity Network Fusion method (Wang et al, Nature Methods, 2014). Stable clusters were defined using spectral clustering and characterised using available clinical data.
Results: Three stable clusters were defined, separated mainly by immune cell composition in sputum. Cluster 3 (C3) is clinically milder, with higher FEV1% predicted and FEV1/FVC ratio. C1 has a more pronounced Th2 phenotype than C2 and C3 as defined by the percentage of sputum eosinophils and the higher periostin levels. C2 regroups the patients with both high sputum neutrophil and eosinophil counts, see table.
| Variables/Clusters (N) | C1 (21) | C2 (21) | C3 (31) | P-value |
| FEV1% predicted | 65 ± 23 | 61 ± 21 | 78 ± 21 | 1.7E-2 |
| FEV1/FVC | 0.58 ± 0.10 | 0.57 ± 0.15 | 0.66 ± 0.11 | 1.4E-2 |
| Neutrophils % | 33 ± 19 | 76 ± 18 | 51 ± 18 | 6.7E-10 |
| Eosinophils % | 31 ± 26 | 7.9 ± 10 | 3.4 ± 5.3 | 3.2E-08 |
| Macrophages % | 34 ± 22 | 16 ± 16 | 44 ± 18 | 5.1E-06 |
| Periostin ng/ml | 57 ± 21 | 56 ± 18 | 42 ± 9 | 7.0E-3 |
Conclusion: We have been able to combine different sputum omics datasets to define stable clusters of patients and characterise them with clinical and biological features. This study may help refining phenotypes of severe asthma.
IMI grant n°115010 (U-BIOPRED).
- Copyright ©ERS 2015
