Characterization of anticancer agents by their growth inhibitory activity and relationships to mechanism of action and structure

An analysis of the growth inhibitory potency of 122 anticancer agents avail able from the National Cancer Institute anticancer drug screen is presented . Methods of singular value decomposition (SVD) were applied to determine t he matrix of distances between all compounds. These SVD-derived dissimilari ty distances were used to cluster compounds that exhibit similar tumor grow th inhibitory activity patterns against 60 human cancer cell lines. Cluster analysis divides the 122 standard agents into 25 statistically distinct gr oups. The first eight groups include structurally diverse compounds with re active functionalities that act as DNA-damaging agents while the remaining 17 groups include compounds that inhibit nucleic acid biosynthesis and mito sis. Examination of the average activity patterns across the 60 tumor cell lines reveals unique 'fingerprints' associated with each group. A diverse s et of structural features are observed for compounds within these groups, w ith frequent occurrences of strong within-group structural similarities. Cl ustering of cell types by their response to the 122 anticancer agents divid es the 60 cell types into 21 groups. The strongest within-panel groupings w ere found for the renal, leukemia and ovarian cell panels. These results co ntribute to the basis for comparisons between log(GI(50)) screening pattern s of the 122 anticancer agents and additional tested compounds.