Stability-indicating method experimental approach

Bursi et al. (2001) reported two methods to calculate the stability of testosterone-like steroids. These were the use of a decision tree and molecular descriptors or quantum mechanical methods. For satisfactory accuracy, Bursi and colleagues (2001) had to use a 3-21G basis set with spin correction and equilibrium geometries. This required 12 hours of computation for reactants and products. Optimization of transition state geometry was also required. The simpler decision tree analysis approach indicated that descriptors such as the volumes and, to a lesser degree, the shape were important. Correlations of calculated and experimental rates of metabolism were reported. 

Thus, a comparison between HPLC-derived lipophilicity indices and calculated log P values for a series of 8-substituted xanthines showed a clear influence of conformational effects. 8 In this case, Rekker s method was unable to take 3D effects into account, but the difference between experimental and predicted values was structure dependent rather than constant. Conformational analyses confirmed that a smaller than predicted lipophilicity was associated with folded conformers stabilized by hydrophobic and van der Waals forces and having part of their nonpolar surface masked from the aqueous phase. A 4D theoretical approach (log P calculations by MLP for conformers generated by high temperature molecular dynamics) suggests that these effects should be lower in an w-octanol/water system than in RP-HPLC. Indeed, the n-octanol/water system is not the most suitable model to study intramolecular interactions in nonpolar media because a surprisingly high proportion of water is dissolved in the w-octanol. Recall, however, that w-octanol, despite some limitations, was selected by many workers in the field as a model for biological membranes. 

---