(446) Metabolites in Early Life Bronchoalveolar Fluid Associate With Future Bronchiectasis Risk in Children With Cystic Fibrosis

Background: Early life biomarkers that predict later structural lung disease (SLD) in children with cystic fibrosis (CF) have the potential to facilitate personalised care, offer insights into disease pathophysiology, and suggest novel therapeutic targets. While elevated neutrophil elastase (NE) is associated with future bronchiectasis (BE), its low predictive power indicates a need for improved biomarkers.

Methods: Biomarkers were analysed in BAL collected annually from 0.5-6 years in 14 AREST-CF participants with no evidence of BE on CT at 9 years old (mildest disease quartile) and compared to BAL from 14 subjects with BE matched by sex, age, genotype, and pancreatic status. Biomarkers included inflammatory markers collected by AREST CF (NE, BAL cell proportions, and IL-8) as well as those from untargeted, discovery metabolomics (positive and negative mode), with a subset of metabolites verified on a targeted platform. Metabolomic analysis focused on metabolites detected on multiple platforms, with values normalized across platforms and averaged. To facilitate analyses, biomarker concentrations in BAL collected between ages 0-2 for each subject were averaged as early preschool, with BAL measurements between ages 3-5 averaged as late preschool. Differences between groups were assessed by T-tests, with predictive ability (area under the curve, AUC) calculated by multiple logistic regression.

Results: In early preschool (0-2), several metabolites detected in untargeted metabolomics and verified in targeted studies were significantly elevated in BAL from subjects who developed BE, including leucine (p=0.039 by cross-platform normalized average), phenylalanine (p=0.037), and lactate (p=0.003). Deoxyguanosine, methionine sulfoxide, and homoserine were elevated in discovery metabolomics (p-values 0.026, 0.010, and 0.039 respectively), though were not included in the targeted panel. NE was elevated in those who developed BE (p=0.040).



In later preschool (3-5), adenosine monophosphate (AMP) was elevated (p < 0.001) and cytidine decreased (p < 0.001) by both targeted and untargeted metabolomics in BAL from those who develop BE. Deoxyguanosine also remained elevated (p=0.009). NE (p=0.019), and % neutrophils (p=0.005) were also associated with BE.



Though elevated, NE was poorly predictive of future BE in the 0-2 age range (AUC 0.52, p=0.87). Adding leucine, phenylalanine, and lactate improved prediction (AUC 0.86, p=0.0023), and adding metabolites detected in discovery metabolomics (deoxyguanosine, methionine sulfoxide, homoserine) further increased predictive power (AUC 0.98, p< 0.0001). In later preschool, multiple individual biomarkers were highly predictive including NE (AUC 0.93, p< 0.0001), cytidine (AUC 0.93, p=0.0001), and deoxyguanosine (AUC 0.91, p=0.0002)

Conclusions: We identified multiple early life biomarkers of future SLD, many of which outperform NE in early preschool age. Several markers validate previous associations with early CF lung disease (e.g., lactate, AMP, and methionine sulfoxide), while others, such as cytidine and deoxyguanosine, may reflect novel pathophysiology. If validated, these biomarkers may help with early identification of children with CF at highest risk for SLD.

Acknowledgements: Support: NHMRC SYNERGY Award APP1183640.

References: