Structure-cytotoxicity relationships for a series of HEPT derivatives

Structure-cytotoxicity relationships were studied for a series of 90 HEPT derivatives by means of multiple linear regression (MLR) and artificial neural network (ANN) techniques. The values of log(1/CC50) (CC50=cytotoxic dose of compound required to reduce the proliferation of normal uninfected MT-4 cells by 50%) of the studied compounds were correlated with the descriptors encoding the chemical structures. Using the pertinent descriptors revealed by the regression analysis, a correlation coefficient of 0.935 ( s=0.149) for the training set ( n=81) was obtained for the ANN model with a 5-6-1 configuration. The results obtained from this study indicate that the cytotoxicity of HEPT derivatives is strongly dependent on hydrophobic factors, mainly log P(R1), and dependent on the steric factors, especially SigmaMW(R3+R4). Comparison of the descriptors' contribution obtained in MLR and ANN analysis shows that the contribution of some of the descriptors to cytotoxicity may be non-linear.