Translating breath-based asthma phenotyping into clinical practice

According to the world health organization, asthma and COPD represent the most common chronic respiratory diseases, with respectively 235 and 65 million of people concerned worldwide. These diseases are not curable. Nevertheless, various forms of treatment that help to dilate major air passages and improve shortness of breath can help to increase the quality of life for patients. However, disease phenotyping relies on invasive sputum analysis. In addition, neutrophilic and eosinophilic inflammation phenotypes have similar symptoms but requires different treatment approaches. To improve monitoring of patients, they have to be better phenotyped. It would allow quick adjustment of the treatment.
In this context, we have been developing a breath-based phenotyping approach using thermal desorption and two-dimensional gas chromatography coupled to high resolution mass spectrometry (TD-GC×GC-HRTOFMS). This method allows patient phenotyping with similar performances than the routine clinical practice (Accuracy 72%, AUROC 0.72). Moreover, the combination of breath with other methods, such as FeNO and blood cell count, significantly increases the model performance (Accuracy 76%, AUROC 87%).
Our current efforts are focused on facilitating the translation of the method to the clinic and to better understand the origin of the markers used for the classification. To facilitate such translation, we have been working on direct mass spectrometry approach using full scan SIFT-MS. This method allows to phenotype patients with an accuracy around 75 % in less than 3 min.
This work is included in the continuous effort around breath research to develop clinical implementation.