Combining crystalline and polymeric excipients in API solid dispersions - Opportunity or risk?

Amorphous solid dispersions (ASDs) are often metastable against crystallization of the active pharmaceutical ingredient (API) and thus might undergo unwanted changes during storage. The crystallization tendency of ASDs is influenced by the API crystallization driving force (CDF) and the mobility of the molecules in the ASD. Low molecular weight-excipients are known to stabilize amorphous APIs in so-called co-amorphous formulations. Due to their success in stabilizing co-amorphous APIs, low-molecular weight excipients might also enhance the stability of polymeric ASDs. In this work, we investigated the potential of combined low-molecular weight excipient/polymer formulations with in-silico tools and validated the predictions with long-term stability tests of the most promising excipient/polymer combinations. The considered critical quality attributes for the ASDs were the occurrence of amorphous phase separation, API CDF, and molecular mobility in the ASD. As an example, carbamazepine/polyvinylpyrrolidone ASDs were investigated combined with the excipients fructose, lactose, sucrose, trehalose, saccharin, tryptophan, and urea. Although all excipients had a negative impact on the ASD stability, saccharin still turned out to be the most promising one. Long-term stability studies with ASDs containing either saccharin or tryptophan verified -in agreement to the predictions- that API crystallization occurred faster than in the reference ASDs without additional excipient. This work showed that the addition of crystalline excipients to polymeric ASDs might not only offer opportunities but might also bear risks for the long-term stability of the ASD, even though the crystalline excipient stabilizes the polymer-free API. Consequently, excipients should be evaluated based on the thermodynamic phase behavior of the individual mixture of API/polymer/excipient, rather than based on pure-component properties of the excipient only. In-silico predictions proposed in this work remarkably decrease the number of screening tests for identifying suitable formulation excipients.