XRPD Quantification of Crystalline Content in Amorphous API within Formulation
Often the objective of our projects is to perform controls of the API state during different drug manufacturing steps.
Beyond this we aim to monitor the state of an API within the finished drug products and to compare it to the original design specifications.
The quantification of low-level polymorph impurities within the formulation of finished drug
products presents some analytical challenges. These challenges relate to the low
concentration of the active pharmaceutical ingredient within the formulation and the
necessity to quantify the minor, few percent-scale impurities of an API, which is biased by heavy placebo contribution.
During the validation of a quantification method, different factors must be considered in line with the regulatory guidelines: repeatability, linearity, range and lowest quantification limit LOQ.
The rationale for the current study was the initial screening results obtained during the
development of a novel cholesterol-lowering formulation. Using differential XRPD analysis,
we found that during the granulation step an amorphous API could, in a certain amount,
convert to a crystalline form. The presence of crystalline forms in an amorphous API, even at very low concentrations, poses a risk of initiating further re-crystallisation and is therefore highly undesirable. The scope of the study therefore became to develop and validate the method for quantificating small amounts of crystalline API forms with the lowest possible LOQ.
We developed a quantification method using the single peak of a differential pattern as a marker.
In this particular case series of samples were prepared with seven different concentrations of crystalline forms within the full range of possible concentrations.
Figure 1. The marker peak of the crystalline form
measured at different concentrations shows an overlap with the placebo peak (on the right). The pattern measured with the unknown formulation is shown by markers.
From measured XRPD patterns, by peak deconvolution we extracted the integral
intensities of the marker peak to build a linear regression curve (Figure 2).
Figure 2. Linear regression line built over the
range of concentrations.
From the regression curve, we obtained the estimate for LOQ equal to 2% wt and for the range and accuracy.
When the method was applied to the quantification of formulations from different batches, the concentration of the crystalline form exhibited a strong correlation with the particular process parameter.
The problem was solved once the change in the process parameter was introduced at the