(T1130-03-17) Kinetic Barriers to Disproportionation of Salts of Weakly Basic Drugs

Purpose: Active pharmaceutical ingredients (APIs) are often prepared as salts to enhance their aqueous solubility, dissolution rate and solid-state properties, as well as to improve chemical stability and product shelf-life. Many weakly basic drug salts tend to disproportionate, wherein the API salt converts to the free base, which can severely impact product quality and patient exposure. The potential for disproportionation in oral solid dosage forms during storage has generated considerable interest in risk assessment approaches for disproportionation. Merritt and coworkers addressed this topic through the development of a model to predict disproportionation based on thermodynamic considerations with input parameters of the free base and salt solubility, pK_a and pH_max.¹ The pH_max is considered a key parameter when considering the risk of a salt undergoing disproportionation; if the microenvironmental pH remains above pH_max then disproportionation will occur, and vice versa. Despite the seemingly straightforward theory describing disproportionation, prediction of disproportionation risk remains challenging. Recent studies found that disproportionation did not occur for some compounds,^2-3 even though there was a thermodynamic driving force for conversion, suggesting that kinetic barriers need to be considered in addition to the thermodynamic descriptors. Herein, we demonstrate that for the salts of some weakly basic compounds, disproportionation is not seen, even for favorable thermodynamic conditions, due to slow kinetics for nucleation of the free base form. Furthermore, the nucleation kinetics depend on the difference between the pH and pH_max, in a compound-specific manner.

Methods: The disproportionation kinetics of three model compounds, including pioglitazone hydrochloride (PIOH), sorafenib tosylate (SORT) and atazanavir sulfate (ATZS), were measured in buffered slurries of known pH values by Raman spectroscopy. Concurrently, a pH meter capable of responding rapidly and recording pH as a function of time, was used to follow the kinetics of conversion. X-ray powder diffraction was employed to confirm the Raman results, evaluating the solids present in the slurry at the end of the experiment.

Results: During disproportionation, the pH vs. time profiles were in sync with the change in the Raman spectra of three compounds in which the pH of the slurries started dropping rapidly once the salt conversion commenced then remained unchanged by the end of the disproportionation process (Figure 1), suggesting that the conversion kinetics could be followed simply by monitoring the pH. The disproportionation kinetics were significantly faster at pH values much higher than pH_max, but decreased dramatically as pH_max was approached, except for ATZS. At high pH values, the disproportionation timeframe substantially differs between compounds: it was only a few minutes for PIOH but increased to about an hour for SORT and to several hours for ATZS (Figure 2). The disproportionation process in a slurry consists of three steps: (i) dissolution of the salt form, (ii) conversion of the salt to the free base, and (iii) crystallization of the free base from supersaturated solution. The difference in the disproportionation kinetics between the three compounds was in line with the variation in the nucleation induction time of the free bases in the solution. At a concentration equal to the amorphous solubility, pioglitazone nucleated instantly whereas it took approximately 2.5 and 5 hours for sorafenib and atazanavir, respectively, to crystallize. Of further note, for PIOH, a compound whose disproportionation is facile in solid blends with basic excipients, there was a small window between pH_max and the pH where disproportionation kinetics became rapid; while the pH region with a deadzone for the conversion was larger for SORT and ATZS, the two compounds previously noted to be resistant to disproportionation in the presence of proton accepting excipients.³ The salt slurry of ATZS in water exhibited a low pH value (around 1.7) within the deadzone where no disproportionation occurred (Figure 3). The pH increased upon addition of basic excipients namely croscarmellose sodium and magnesium stearate, and then decreased as the salt released protons upon free base formation. Notably, croscarmellose sodium increased the pH and thus accelerated disproportionation much more rapidly than magnesium stearate, as it has a higher affinity for water, and hence proton transfer is expedited. Disproportionation ceased once the pH dropped into the deadzone identified in Figure 2.

Conclusion: The connection between the deadzone pH range with hindered disproportionation kinetics and the dynamic pH profile induced by addition of an excipient helps rationalize why disproportionation in solid formulations containing basic excipients is not always observed even though predicted based on thermodynamic considerations. In other words, the excipient may not drive the pH into the critical pH region where the phase transformation kinetics are favored. This research provides an experimental approach to understand if there are kinetic barriers to disproportionation, which in combination with the Merritt model can be used to better understand the risk for salt disproportionation in a formulation. Furthermore, it confirms that a solution-based thermodynamic model is mechanistically appropriate to predict disproportionation, when combined with a consideration of kinetic factors.

References: 1. Merritt, J.; Viswanath, S.; Stephenson, G. Implementing Quality by Design in Pharmaceutical Salt Selection: A Modeling Approach to Understanding Disproportionation. Pharmaceutical Research 2013, 30 (1), 203-217.
2. Patel, M. A.; Luthra, S.; Shamblin, S. L.; Arora, K.; Krzyzaniak, J. F.; Taylor, L. S. Impact of Solid-State Form on the Disproportionation of Miconazole Mesylate. Molecular Pharmaceutics 2018, 15 (1), 40-52.
3. Patel, M. A.; Luthra, S.; Shamblin, S. L.; Arora, K. K.; Krzyzaniak, J. F.; Taylor, L. S. Assessing the Risk of Salt Disproportionation Using Crystal Structure and Surface Topography Analysis. Crystal Growth & Design 2018, 18 (11), 7027-7040.


Figure 1. (A) Raman spectra and (B) pH profile of SORT (26.7 mg/mL) as a function of time during disproportionation at pH 9.0. Disproportionation commenced at 30 min with a decrease in intensity of Raman peaks at 1691 and 1716 cm-1 whereas the peak at 1701 cm-1 emerged simultaneously with the rapid drop in slurry pH from 9.0 to approximately 8.1. Both the Raman spectra and pH remained virtually unchanged after 60 min, indicating the end of the disproportionation process.


Figure 2. Disproportionation duration as a function of pH for three model compounds showing deadzone pH ranges with hindered disproportionation kinetics. The dashed lines on the left indicated the pHmax values of the APIs (1.0; 0.7; and 0.9 for PIO, SOR and ATZ, respectively).


Figure 3. pH vs. time profiles of ATZS (26.7 mg/mL) in water upon addition of various amounts of (A) croscarmellose sodium (CCS) or (B) magnesium stearate (MgSt); and (C) Disproportionation time as a function of excipient slurry concentration.