(T1430-02-12) Novel Melting Point Depression Method to Predict Drug Dissolution during Melt Extrusion of PLGA Millifiber Implant

Purpose: Solubility curves for small molecule drugs in polymeric drug delivery matrices may have great value when designing hot melt extrusion (HME) processes, as drug dissolution into the polymer melt may impact melt rheology, drug release and product stability. Drug-polymer solubility is often determined by Differential Scanning Calorimetry (DSC) melting point depression measurements of drug/polymer blends^1,2. The accuracy of these measurements depends on small particle sizes of drug and polymer, as rapid mixing at the equilibrium solubility temperature requires high interfacial area between particles. One preferred method of particle size reduction is cryo-milling. In this study, we developed a novel melting point depression method for a small molecule drug in molten poly(lactic-co-glycolic acid) (PLGA), wherein high interfacial area of drug and polymer is achieved by crystallization of an amorphous solid dispersion (ASD) rather than cryo-milling. This method was generally predictive of actual dissolved drug amounts in extrudate samples.

Methods: 50:50 PLGA (acid-terminated, inherent viscosity: ~0.2 dL/g, T_g: 47 °C) was jet-milled to reduce polymer filaments to spherical particles. Crystalline Drug X (basic p<em>K_a: 1.9, LogP: 1.8) was used as received from the manufacturer. Melting point depression: drug/polymer powder blends were prepared at drug loadings of 20, 40, 60, and 80% w/w by geometric mixing in a benchtop mill, then loaded into DSC pans with a pinhole opening. DSC samples were subjected to a heat-cool-heat procedure: The blend was heated to 20 °C above the pure drug’s melting point to allow mixing of drug and polymer by passive diffusion; The molten mixture was rapidly cooled to form a homogenous ASD (melt quenching); The sample was crystallized and then remelted in a slow heating scan. Multiple heating rates were used for the initial melt (step 1) and second melt (step 3), and melting data were extrapolated to a heating rate of 0 °C/min. In a second experiment, melt-quenched ASDs were crystallized by an isothermal hold at 50 °C/0% RH for 7 days. Glass transition temperature (T_g) calibration: T_gs of melt-quenched ASDs were measured by a modulated scan and fit to the Gordon-Taylor equation. T_gs of extrudates were measured by a single modulated scan. Melt extrusion: Powder blends of 40 and 60% drug loading were fed into a benchtop twin screw extruder with a barrel temperature setting of 100 °C. Extrudates were pulled into millifibers onto a guide plate set to 45 °C, then stored at 5 °C until DSC analysis.

Results: Melting point depression could not be detected in the initial powder blends (Fig. 1a). However, melt enthalpy depression was evident, indicating exothermicity of drug dissolution (Fig. 2a). Kinetic factors such as particle size and heating rate may not compromise the measurement of melt enthalpy depression if exothermic drug-polymer interactions form quickly enough after drug melting for exothermic and endothermic signals to overlap in the DSC signal; this may be the case in our low-viscosity drug/PLGA melt. Melt enthalpy approached nil at 20% drug loading, suggesting low chemical potential for crystallization below 20% loading. This fact, combined with the difficulty of detecting broad and low-intensity depressed melting points, likely precludes melting point measurements below 20% loading. Melting point depression was significant in the crystallized ASDs (Figs. 1b, 1c). The 40% loading ASD did not crystallize by the slow heating method, likely due to longer nucleation induction times and slower crystal growth kinetics at lower drug loading. The 40% loading ASD completely crystallized by the annealing method, however, as indicated by the comparable melt enthalpy of the annealed samples and initial powder samples (Fig. 2a). Some replicates of neat drug in the slow heating method showed melting points of 90 °C and 124 °C, consistent with a known polymorph (data not shown). However, all nine replicates of annealed neat drug had melting points matching the initial powder, suggesting no polymorph conversion by the annealing method.Melting point curves were extrapolated to low drug loadings (Fig. 2b). Extrapolation predicted melt peak temperatures of 107 °C at 20% drug loading. It is possible our actual extrusion temperature was near 107 °C due to viscous heating. Further study is needed to determine the actual extrusion temperature and if melting onset, peak, or endset is most informative for our system. Finally, the T_gs of ASDs and extrudates were compared (Fig. 3). All samples showed a single T_g, indicating a homogenous amorphous phase. By calibration, the amorphous fraction of all extrudates was approximately 20% drug.

Conclusion: Melting point depression was easier to measure in crystallized ASDs than powder blends of drug and polymer. We posit that this is due to high interfacial area of small drug crystals and polymer following crystallization of a homogenous, molecular-level dispersion. This method may be useful for formulation development of drug/polymer systems for which the ASD has a relatively low T_g and can crystallize in a timely manner.

References: 1. Marsac, P.J., Shamblin, S.L., Taylor, L.S., 2006. Theoretical and practical approaches for prediction of drug-polymer miscibility and solubility. Pharm. Res. 23, 2417–2426.
2. Tao, J., Sun, Y., Zhang, G.G.Z., Yu, L., 2009. Solubility of small-molecule crystals in polymers: D-Mannitol in PVP, indomethacin in PVP/VA, and nifedipine in PVP/VA. Pharm. Res. 26, 855–864.

Fig. 1: DSC thermograms for melt of powder blends (a); crystallization and melt of ASDs (b); and melt of ASDs crystallized by isothermal annealing (c). Data shown were collected at a heating rate of 5 °C/min.

Fig 2: Melt enthalpy as a function of drug loading for the powder blends and annealed ASDs (a); and melting temperature as a function of drug loading for the annealed ASDs (b).

Fig. 3: Tgs of ASDs fit to the Gordon-Taylor equation and Tgs of extrudate samples.