Introduction Xe MRI and PFTs provide complex information about lung physiology which may allow improved patient phenotyping.

Objective To investigate possible data driven phenotypes of obstruction based on Xe MRI and lung physiology using cluster analysis.

Methods Patients with asthma and/or COPD taking part in the NOVELTY study [NCT02760329] were recruited from primary care and assessed post-bronchodilator. K-means clustering was performed on 10 metrics derived from Xe MRI (ventilation, acinar dimensions and gas transfer) and PFTs (spirometry, body plethysmography and gas transfer). Inter-cluster analysis on clinical outcomes was then performed.

Results 148 patients, aged 28?82 years, with asthma (73), asthma+COPD (50) or COPD (25) were grouped into 3 clusters (C). There were significant differences between all clusters for 9/10 of the MRI and PFT metrics.

C1 (n=24) had the most disease measured by Xe MRI and PFTs. 54% had COPD and 42% had asthma+COPD. 96% were ever smokers.

C2 (n=67) had mild lung physiology on MRI and PFTs (69% had normal PFTs) and were the youngest. 84% had asthma and 15% had asthma+COPD.

C3 (n=57) MRI and PFT metrics were poorer than cluster 2 and better than cluster 1. 53% had asthma+COPD and 28% had asthma.

Inter-cluster analysis; C1 had more exacerbations over the previous 3 years, more symptoms (RSQ, CAAT), lower quality of life (SGRQ) and more neutrophils than C2 and C3.

Conclusions Three data driven clusters of obstructive lung disease severity were identified based on MRI and PFT measurements and independent of clinical diagnosis which link to clinical outcomes. These clusters may therefore help to predict worsening quality of life and future exacerbations.