B102-03 Emphysema Cluster Size Is Associated With Perturbation of the Leptin-ghrelin Axis

Rationale Quantitative emphysema on chest computed tomography (CT) is commonly studied in chronic obstructive pulmonary disease (COPD). The fractal dimension D of emphysema cluster size measures the spatial homogeneity and size of emphysema clusters and is reduced in people with larger emphysema cluster sizes, with a value less than 1.05 associated with COPD. Little is known about the systemic molecular effects of emphysema cluster size. We designed this study to discover protein biomarkers of low D. Methods We analyzed participant data from the Genetic Epidemiology of COPD (COPDGene) study. Quantitative CT analysis was performed on 9,682 COPDGene participants at Phase 1 visit. Lung segmentation was derived using an in-house deep learning-based algorithm. Mild emphysematous regions were identified using a standard low attenuation threshold of -910 HU. The D910 parameter was calculated as the negative slope of the power-law relationship describing emphysema cluster size distribution. Three-dimensional connected component analysis with 26-connectivity was employed to identify all connected emphysematous clusters. Proteins were measured using the SomaScan 1.3k platform on plasma samples from 1,126 participants at Phase 1 visit. All participants with available CT and protein data were included and were grouped into “low D910” (D910<1.05) and “normal D910” (D910≥1.05) groups. Protein analysis was conducted using log-count-per-million-transformed raw protein counts via limma-voom after adjusting for age, sex, percent emphysema, percent predicted FEV1 (FEV1pp) by GLI Global equation, and pack-years of smoking. Gene set enrichment analysis was performed using FUMA. Results A total of 1,003 participants were included in this analysis. Participants with low D910 were older (63.5±9.0 vs. 60.9±9.3 years, p < 0.001), more likely to be Black or African American (12% vs. 8%, p = 0.03), had higher smoking exposure (49.0±23.8 vs. 39.8±25.6 pack-years, p < 0.001), had worse spirometric obstruction (FEV1pp 66.7±28.8 vs. 90.3±21.2, p < 0.001), and were more likely to have COPD (67% vs. 31%, p < 0.001). After correction for co-variates, low D910 was associated with differential abundance of 40 plasma proteins (Figure 1). Leptin was lower in participants with low D910 (log2 fold-change(logFC)=-0.70, adjusted p-value(padj)=4.26x10-12), as was growth hormone receptor (logFC=-0.29, padj=2.12x10-8). The BioCarta gene set for the ghrelin pathway was significantly enriched in positively-associated proteins (padj=5.83x10-3) and insulin-like growth factor binding protein-1 (IGFBP1) showed the strongest positive association with low D910 (logFC=0.54, padj=4.58x10-7). Conclusions We have identified emphysema cluster size as a trait that is linked to plasma markers of body composition and metabolism. Further analysis is needed to understand the physiological basis of this relationship.