Severe asthma and personalized approach in the choice of biologic

Purpose of review: Severe asthma requires intensive pharmacological treatment to achieve disease control. Oral corticosteroids are effective, but their use is burdened with important side effects. Biologics targeting the specific inflammatory pathways underpinning the disease have been shown to be effective but not all patients respond equally well. As we treat more patients than those who can respond, our inability to predict responders has important healthcare costs considering that biologics are expensive drugs. Thus, a more precise choice of the 'right patients' to be prescribed with the 'right biologics' would be desirable. Recent findings: Machine learning techniques showed that it is possible to increase our ability to predict outcomes in patients treated with biologics. Recently, we identified by cluster analysis four different clusters within the T2 high phenotype with differential benralizumab response. Two of these clusters, characterized by higher levels of inflammatory markers, showed the highest response rate (80-90%). Summary: Machine learning holds promise for asthma research enabling us to predict which patients will respond to which drug. These techniques can facilitate the diagnostic workflow and increase the chance of selecting the more appropriate treatment option for the individual patient, enhancing patient care and satisfaction.