Introduction: Continuing inhaled corticosteroid (ICS) use does not benefit all patients with COPD, yet it is difficult to determine which patients may safely sustain ICS withdrawal. Although eosinophil levels can facilitate this decision, better biomarkers could improve personalised treatment decisions.
Methods: We performed transcriptional profiling of sputum to explore the molecular biology and compared the predictive value of an unbiased gene signature versus sputum eosinophils for exacerbations after ICS withdrawal in COPD patients. RNA-sequencing data of induced sputum samples from 43 COPD patients were associated with the time to exacerbation after ICS withdrawal. Expression profiles of differentially expressed genes were summarised to create gene signatures. In addition, we built a Bayesian network model to determine coregulatory networks related to the onset of COPD exacerbations after ICS withdrawal.
Results: In multivariate analyses, we identified a gene signature (LGALS12, ALOX15, CLC, IL1RL1, CD24, EMR4P) associated with the time to first exacerbation after ICS withdrawal. The addition of this gene signature to a multiple Cox regression model explained more variance of time to exacerbations compared to a model using sputum eosinophils. The gene signature correlated with sputum eosinophil as well as macrophage cell counts. The Bayesian network model identified three coregulatory gene networks as well as sex to be related to an early versus late/nonexacerbation phenotype.
Conclusion: We identified a sputum gene expression signature that exhibited a higher predictive value for predicting COPD exacerbations after ICS withdrawal than sputum eosinophilia. Future studies should investigate the utility of this signature, which might enhance personalised ICS treatment in COPD patients.