“Predicting” Crystal Forms of Pharmaceuticals Using Hydrogen Bond Propensities: Two Test Cases

The Cambridge Structural Database
incorporates an algorithm, based
on data mining and statistical analysis of the structures in the database,
for evaluating the “risk of polymorphism” in any compound.
We have applied that algorithm to a number of pharmaceutically important
compounds from the European Pharmacopoeia for which multiple crystal
forms have not been reported. The survey suggested the possibility
of polymorphism in at least two compounds, bufexamac and meglumine.
The statistical analysis and the subsequent experimental search for
polymorphism are reported here. While polymorphism was detected and
characterized in both cases, the structural results were not necessarily
compatible with the basis for the measure of the “risk of polymorphism”.