Mucus Plug Lobar Distribution in COPD Using Automatic AI-Based Mucus Plug Quantification

<bold>Aims:</bold> In COPD mucus plugs (MPs) can be visually assessed on chest CT scans, however it is time consuming and requires training. Large scale investigations of MPs phenotypes with respect to MPs incidence, extent, and lobar distribution were therefore not feasible. We used a fully automated AI-based MPs analysis to count the number and volume MPs and determine their lobar distribution in (ex)smokers with COPD across different GOLD-stages, and in never smoker controls. <bold>Methods:</bold> All subjects from Phase 1 of the COPDGene study with analysable inspiratory CT scans were included, consisting of 9521 (ex)smokers and 107 never smoker controls. <bold>Analysis:</bold> Fully automatic quantification of MPs assessing plug count, volume, and upper/lower lobe distribution using an AI-based lung quantification platform (LungQ, Thirona, The Netherlands). <bold>Results:</bold> MPs incidence increased with GOLD stage, from 9% in GOLD 0 to 74% in GOLD 4. A lower lobe predominance was found through all GOLD stages. The number and volume of MPs in the only lower lobes group were higher compared to the only upper lobes group (see Table). <bold>Conclusion:</bold> The number of MPs is higher for higher GOLD stages. There is a predominance of MPs in the lower lobes which might result from gravitational effects, potentially more structural airway changes and impaired mucociliary clearance in the lower lobes, micro-aspirations and subsequent higher local inflammation.