Statistical approaches for dissolution profile comparisons of metformin film-coated tablets

The objective of this work was to apply several statistical approaches to profile comparison on
dissolution data of Metformin immediate release film-coated tablets to assure that the developed
formulation of the test product that is similar to the reference product (Glucophage film-coated
tablets). The evaluated medicinal product belongs to BCS Class III (high solubility, low
permeability). The similarity testing of the dissolution profile was performed on the highest strength
of the dosage form, following the regulatory requirements for bioequivalence study. The obtained
results showed that the simple standard difference and similarity (f1 and f2)-factors are not applicable
and therefore practically not useful for comparison of dissolution profiles because one of the
regulatory requirements that the relative standard deviation or coefficient of variation of any product
should be less than 20% for the first point and less than 10% from second to last time point, which
was not achieved in the present case. Therefore, more advanced multivariate methods such as modeldependent approaches coupled to multivariate statistics, the multivariate model-independent
approach based on generalized statistical distance (such as e.g. Mahalanobis distance) have been
applied for evaluation of dissolution profiles. This approach appeared to provide extra arguments
when deciding if two profiles are similar, as it allows for a better description of the dissolution
processes (in the sense of e.g. the rate and amount of the active substance dissolved) and thus a
better prediction of in vivo performance. Comparison with traditional methods such as those based
on fit-factors is made, and their shortcomings pinpointed out. Dissolution profiles can be tested for
differences in both level and shape by multivariate-based methods and these methods provide
detailed information about dissolution data, which can be useful also in formulation development to
achieve the final goal – match the actual performance of the released test product to the performance
of a target reference product.