Amorphous Drug-Polymer Salts: Maximizing Proton Transfer to Enhance Stability and Release

Abstract

An amorphous drug-polymer salt (ADPS) can be remarkably stable against crystallization at high temperature and humidity (e.g., 40°C/75% RH) and provide fast release. Here, we report that process conditions strongly influence the degree of proton transfer (salt formation) between a drug and a polymer and in turn the product's stability and release. For lumefantrine (LMF) formulated with poly(acrylic acid) (PAA), we first show that the amorphous materials prepared by slurry conversion and antisolvent precipitation produce a single trend in which the degree of drug protonation increases with PAA concentration from 0% for pure LMF to ∼100% above 70 wt % PAA, independent of PAA's molecular weight (1.8, 450, and 4000 kg/mol). This profile describes the equilibrium for salt formation and can be modeled as a chemical equilibrium in which the basic molecules compete for the acidic groups on the polymer chain. Relative to this equilibrium, the literature methods of hot-melt extrusion (HME) and rotary evaporation (RE) reached much lower degrees of salt formation. For example, at 40 wt % drug loading, HME reached 5% salt formation and RE 15%, both well below the equilibrium value of 85%. This is noteworthy given the common use of HME and RE in manufacturing amorphous formulations, indicating a need for careful control of process conditions to ensure the full interaction between the drug and the polymer. This need arises due to the low mobility of macromolecules and the mutual hindrance of adjacent reaction sites. We find that a high degree of salt formation enhances drug stability and release. For example, at 50% drug loading, an HME-like formulation with 19% salt formation crystallized faster and released only 20% of the drug relative to a slurry-prepared formulation with 70% salt formation. Based on this work, we recommend slurry conversion as the method for preparing ADPS for its ability to enhance salt formation and continuously adjust drug loading. While this work focused on salt formation, the impact of process conditions on the molecular-level interactions between a drug and a polymer is likely a general issue for amorphous solid dispersions, with consequences on product stability and drug release.

Keywords: amorphous drug−polymer salt; dissolution; lumefantrine; physical stability; poly(acrylic acid); tropical conditions.

Scheme 1. Structures of Lumefantrine (LMF), Clofazimine (CFZ), and Poly(acrylic acid) (PAA)

Figure 1

Typical XPS N spectra of amorphous LMF-PAA. These materials were prepared by slurry conversion using PAA 450 kg/mol. With increasing PAA concentration (decreasing drug loading), the unprotonated N peak decreases and the protonated N peak increases. The curves are Gaussian fits of the peaks.

Figure 2

Surface concentration of LMF as a function of bulk concentration. The diagonal line indicates perfect agreement of the two concentrations.

Figure 3

Protonated fraction of LMF in amorphous formulations with PAA of different M_Ws. For PAA 450 and 4000 kg/mol, some results were obtained with more vigorous mixing. A single trend is observed regardless of PAA M_W, indicating the reaction had reached equilibrium. The vertical line at w₀ = 88 wt % corresponds to one LMF molecule (M_W = 528.9 g/mol) per PAA monomer (M_W = 72.1 g/mol). The curve is a fit to a reaction model (see below).

Figure 4

Protonated fraction of LMF in amorphous LMF-PAA prepared with PAA 4000 kg/mol using the standard slurry method (open symbols) and with enhanced mixing (solid symbols). The latter shows higher and tighter degrees of protonation due to more complete reaction. The curve is a guide to the eye.

Figure 5

Protonated fraction of LMF vs drug concentration. The materials were prepared by slurry conversion (solid symbols) and antisolvent precipitation (open symbols) using PAA of different M_Ws, which are not distinguished. Within experimental error, the materials prepared by the two methods form a single trend. The vertical line at w₀ has the same meaning as that in Figure 3. The curve is a fit to a reaction model (see below).

Figure 6

Comparison of protonation profiles in amorphous LMF formulated with PAA 450 kg/mol by different methods. At the same drug loading, slurry conversion (solid diamonds) achieved more complete salt formation than HME and RE used by Song et al. and a melt-quench method used in this work. The curve through the slurry data is a fit to a reaction model (see below).

Figure 7

Effective equilibrium constant K_eff for the proton transfer between LMF and PAA (eq 3) vs the total AA monomer mole fraction (neutral and deprotonated) x_a0. K_eff increases exponentially with x_a0 (curve).

Figure 8

(a) XPS spectra of two amorphous LMF formulations prepared by slurry conversion and melt quench. Both were prepared with PAA M_W 450 kg/mol at 50% drug loading but had different degrees of salt formation. (b) Stability of the two formulations at (a) 40 °C and 75% RH. The melt-quenched formulation crystallized faster than the slurry-prepared formulation. (c) Dissolution curves of the two formulations in (a) and crystalline LMF. For each sample, the different symbols indicate the results on independent batches. The melt-quenched material reached higher concentration than the crystals, but much lower value than the slurry-prepared material.

Figure 9

Fraction of LMF molecules in amorphous LMF-PAA that are protonated as a function of PAA M_W. For this comparison, the drug loading was fixed at 75%. Maleic acid, a dicarboxylic acid, is used to mimic a PAA dimer. For the polymers, the ability to protonate LMF is insensitive to M_W (horizontal line), while the protonating power could increase for oligomers (sloping line).

Figure 10

Comparison of the degrees of salt formation in LMF-PAA and in CFZ-PAA prepared with PAA 450 kg/mol. At the same drug loading, salt formation is more complete for CFZ than LMF. The vertical line at w₀ corresponds to one drug molecule per PAA monomer (M_W = 72.06 g/mol) with w₀ = 88 wt % for LMF (M_W = 528.9 g/mol) and 87 % for CFZ (M_W = 473.4 g/mol), indistinguishable at the scale used.