Development of QSAR's based on high performance liquid chromatography capacity factors to describe non-specific toxicity

The high performance liquid chromatography (HPLC) capacity factors (k') of a diverse set of lipophilic compounds were measured using several stationary phases. Quantitative structure activity relationships (QSAR's) relating these capacity factors and non-specific toxicity to a mixed marine bacterial culture were developed which indicated that C-18 followed by Silica stationary phases were generally the most effective at modelling toxicity. The use of a polar and non-polar stationary phase connected in series, improved the quality of the simple linear regression equations (SLR's) compared to the SLR's for either C-18 or Amino stationary phases. Second order polynomial regression equations further improved the modelling capability but were not as successful as multiple linear regression equations (MLR's). The form toxicity data was expressed in also influenced the modelling capacity of the QSAR's. Comparison of the QSAR's derived with QSAR's using other physicochemical parameters and molecular descriptors including the octanol-water partition coefficient revealed the log k' QSAR's to be highly suited to the modelling of non-specific toxicity.