Hierarchical clustering analyses may identify potential patterns of marker expression and biologically relevant groupings. As an example, PDGF-AA and PDGF-BB are tightly clustered at baseline, indicating that if a patient had high PDGF-AA levels, PDGF-BB levels also tended to be high. This relationship is maintained at later time points (C2D1 and C3D1), as the clustering of PDGFs reflects similar patterns of change on-treatment.