Selection of Polychlorinated Biphenyls for use in Quantitative Structure-Activity Modelling

Abstract By characterizing the 154 tetra- through heptachlorinated biphenyl (PCB) congeners with a multitude of physico-chemical descriptors, a model representing chemical similarities and differences is achieved. The multivariate characterization of the PCBs was based on 47 physico-chemical descriptor variables, which were summarised by using principal component analysis (PCA). By applying statistical design to the orthogonal scores from the PCA, a 2(4)-factorial design was used to select a set of 16 congeners. In addition, four congeners were added to provide information about the interior region of the chemical domain of PCBs. This set of 20 structurally different congeners is suggested to be used in future quantitative structure-activity relationships (QSARs) for screening of the toxicological and biochemical effects of the PCBs.