B101-06 COPD Spirometric Trajectory Clusters Reveal Differences in the Complement Pathway

Abstract Rationale The progression of lung function decline in chronic obstructive pulmonary disease (COPD) depends on smoking history, age, comorbidities, genomics, and many other factors that inter-depend on one another. Disentangling these complex interactions will give us a better understanding of how COPD develops and improve disease outcome. To this end, our study hypothesizes that there are overarching patterns in lung function decline that depend on the proteome. Methods A total of 6,198 individuals from the COPDGene cohort with at least two spirometric measurements were selected. To quantify differences between spirometric trajectories, we used the Dynamic Time Warping (DTW) similarity. Three variables were used to define the coordinates of each trajectory: FEV1 % predicted, FEV1/FVC, and age. A complete graph was constructed where edge weights represented pairwise DTW similarities and a minimum spanning tree was extracted. The Louvain algorithm was applied to identify communities within the tree. At the 5-year visit, blood samples were taken from COPDGene participants to determine protein expression profiles using the SomaLogic SomaScan® version 4.0 (5.0K) assay. Relative protein abundances were fit to two linear regression models labeled as “reduced” and “full”. The reduced model regressed proteins against age, sex, race, smoking status, platelet count, white blood cell count, and clinical center whereas the full model also included trajectories as a predictor. An ANOVA F-test was performed to determine if the trajectories significantly improved model prediction. Results The Louvain algorithm identified 30 clusters (Fig. 1). Trajectories grouped along the minimum spanning tree with older, stable individuals in the upper left and fast declining, young individuals to the right. The exception was trajectory 30 which rapidly shifts from GOLD 0 to GOLD 2. A functional enrichment analysis of the proteome revealed significant proteins for angiogenesis (VEGF, Ephrin-B2, EphB4), RAS signaling, PI3K-ATK signaling, and the complement pathway (CRP, C9). Looking further into the complement pathway, we saw significant differences between neighboring trajectories 10/30 and 20/28. Trajectories 10 and 28 have comparatively more GOLD 0 individuals initially and show less decline than their counterparts. Conclusion In this study, we investigated how individuals clustered based on their lung function decline and showed how these trajectories relate back to their protein expression. We found that despite trajectories similarities, significant differences in the proteome could be identified. In particular, the complement pathway showed differences in a rapidly declining trajectory. This abstract is funded by: This work was supported by NHLBI R01 HL151421, R01 HL137995, R01 HL178032, and R01 HL159805. The COPDGene study (NCT00608764) is supported by grants from the NHLBI (U01HL089897 and U01HL089856), by NIH contract 75N92023D00011, and by the COPD Foundation through contributions made to an Industry Advisory Committee that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer-Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer and Sunovion.