Spray drying OZ439 nanoparticles to form stable, water-dispersible powders for oral malaria therapy

Nanoparticle formation and characterization

Nanoparticles stabilized by HPMCAS and containing OZ439:oleate were formed via FNP. The FNP process has been described in detail previously [12, 13]. It involves two components: (1) rapid micromixing between a water-miscible organic solvent stream and an aqueous anti-solvent stream, and (2) kinetically arrested aggregation of the drug nanoparticle by adsorption of the stabilizer on its surface. The drug and stabilizing polymer are dissolved in the solvent stream. Upon mixing, which occurs on time scales of O(1) ms, the drug and amphiphilic portions of the stabilizing polymer adsorb on the growing aggregate and arrest growth. Nanoparticles from 25 to 450 nm can be produced with narrow size distributions and at high loadings.

OZ439 is a synthetic trioxolane which was provided in a mesylate salt form (Fig. 1). In the mesylate salt form or free base form, the solubility of OZ439 is too high to create stable nanoparticles by antisolvent precipitation. When either of these forms is used, NPs initially formed during FNP rapidly succumb to Ostwald ripening and grow in size [14, 15]. To form stable NPs, sodium oleate was included in the organic feed stream and acted as a hydrophobic ion pairing agent. Cationic OZ439 and anionic oleate ions paired together, and the resulting complex was sufficiently hydrophobic to precipitate during the mixing step.

Previously, we had applied FNP to OZ439 using a two-inlet lab-scale CIJ mixer [11], which requires a quenching step to stabilize the NPs against Ostwald ripening. As the process is intended to be continuous and at large scale, we here employed a multi-inlet vortex mixer (MIVM) for the formation of nanoparticles. The MIVM allows unequal volumetric flow rates between its four inlets. By introducing three water antisolvent streams, each at three times the volumetric flow rate of the sole organic stream, the MIVM achieved the same final nanoparticle quenching by dilution of the organic solvent concentration, and thus bypassed the quenching step. Figure 2 is a schematic of the two mixers as applied to this process.

Nanoparticles were produced via FNP in the MIVM using sodium oleate as a hydrophobic counterion. OZ439 mesylate (5 mg/mL), sodium oleate (5.38 mg/mL), and HPMCAS 126 (5 mg/mL) were dissolved in a mixture of 33% methanol and 67% THF. This stream was loaded into a syringe and attached to the MIVM, along with three syringes containing DI water. Using a syringe pump (Harvard Apparatus, Massachusetts, USA), the organic stream and water streams were fed into the MIVM at controlled flow rates. The organic stream was fed at 16 mL/min, and each of the water streams was fed at 48 mL/min, such that the resulting NP suspension contained 10% organic solvent by volume.

Nanoparticle mean size, size distribution, and polydispersity were measured by dynamic light scattering (DLS) in a Malvern Zetasizer Nano (Malvern Instruments, Worcestershire, United Kingdom). Following formation, nanoparticle samples were diluted tenfold in DI water immediately prior to measurement to reduce multiple scattering. The Zetasizer was operated at room temperature and used a detection angle of 173°. Measurements were taken in triplicate. DLS data were processed with Malvern’s software using a distribution analysis based on a cumulant model. The cumulant analysis is defined in International Organization for Standardization (ISO) standard document 13321. The calculations of PDI are defined in the ISO standard document 13321:1996 E.

Lyophilization conditions

In order to process nanoparticle suspensions into dry powders for long-term storage and ease of shipping, a drying unit operation like lyophilization or spray drying was required. In lyophilization, a frozen sample is subjected to low temperatures and pressures, and ice and frozen organic solvents are removed by sublimation. Nanoparticles in the suspension are preserved during the freezing process through the addition of a cryoprotectant, usually an inert species that sterically prevents particle–particle interactions, overlap, and aggregation.

The lyophilization protocol used herein was the one optimized in our previous study [11]. In brief, HPMC E3 was added to nanoparticle suspensions following FNP at a 1:1 HPMC E3:solids ratio. The E3 acted as a cryoprotectant as the nanoparticle suspension was immersed in a bath of dry ice and acetone (− 78 °C) and rapidly frozen. Frozen samples were then transferred to a − 80 °C freezer overnight. Lyophilization took place in a VirTis AdVantage Pro BenchTop Freeze Dryer (SP Scientific, Pennsylvania, USA) at − 20 °C under vacuum.

Spray drying conditions

Spray drying was performed using a similar protocol to the one described in Feng et al. [16]. In brief, following nanoparticle formation, HPMC E3 was added to the nanoparticle suspension at a 1:1 HPMC E3:mass ratio to prevent particle aggregation during the drying process. Next, the suspension was fed into a Büchi B-290 spray drier (Büchi Corp., Delaware, USA) via a peristaltic pump at a flow rate of 8 mL/min. Drying parameters such as inlet temperature, mass ratio of added HPMC E3, and aspirator gas flow rate were optimized. The optimal inlet temperature was found to be 145 °C. Following drying, powders were collected and weighed in order to calculate the yield efficiency (YE) of the process. The powder particle size was observed using an Eclipse E200 bright-field microscope (Nikon Instruments, Japan).

Powder characterization: X-ray powder diffraction (XRPD), differential scanning calorimetry (DSC), and water content

XRPD: A D8 Advance diffractometer (Bruker Corporation, Massachusetts, USA) with Ag Kα radiation (λ = 0.56 Å) and a LynxEye-Xe detector was used for XRPD. A polyimide capillary tube (inner diameter = 1 mm) was loaded with 5–10 mg of powder and sealed with quick-setting epoxy. Scattering data were collected over values of 2θ from 3 to 20°, which correspond to Cu Kα 2θ values from 8.2 to 57.0°. A step size of 0.025° (0.067° for Cu Kα radiation) and a rate of 5 s/step were used. Note that in the following sections, all the XRPD results are presented in momentum transfer Q, where Q is a function of wavelength λ and diffraction angle θ \(\left( {Q = \frac{4 \cdot \pi \cdot \sin \left( \theta \right)}{\lambda }} \right)\).

DSC A Q200 DSC (TA Instruments, Delaware, USA) was used for DSC measurements. 5–10 mg of sample was weighed into a pan and equilibrated at 20 °C under dry N₂ atmosphere (50 mL/min). The samples were then heated at 5 °C/min from 20 to 300 °C. The scan was analyzed by TA Instruments Universal Analysis 2000 software.

Water content A V20S Compact Volumetric KF Titrator (Mettler Toledo, Ohio, USA) was used to measure the water content of spray dried powders. 20–30 mg of powder was weighed and then deposited into the device’s titration chamber. After 5 min of stirring, the automatic titration process was performed. Aquastar Titrant 5 and Aquastar Combimethanol (EMD Millipore, Massachusetts, USA) were used as titrants with two-component reagents and solvent, respectively.

OZ439 dissolution

The in vitro solubilization of OZ439 from nanoparticle powders over time in simulated biorelevant media was measured for comparison against unencapsulated OZ439 mesylate. The solubilization protocol was designed to mimic the intended conditions of oral pediatric administration in the developing world; namely, that a mother would add water to the nanoparticle powder before feeding the suspension to an infant.

25 mg of powder, containing 3.37 mg OZ439, was weighed into a scintillation vial. 0.515 mL of water was added, and the powder was allowed to redisperse for 15 min (Step 1, Fig. 3). 0.057 mL of concentrated simulated gastric fluid (FaSSGF) was then added, such that the resulting mixture was at the proper pH and salt concentration of gastric fluid, and the suspension was placed in a water bath at 37 °C (Step 2, Fig. 3). After 15 min, 5.72 mL of either fasted-state (FaSSIF) or fed-state (FeSSIF) simulated intestinal fluid was added to the suspension (Step 3, Fig. 3). Thus the total amount of fluid added was 6.29 mL, and the maximum concentration of solubilized OZ439 was approximately 0.535 mg/mL. It should be noted that during long-term stability studies, the maximum possible concentration of OZ439 in a 25 mg powder sample was lowered slightly due to the sample having absorbed water over time; this was accounted for when calculating percent solubilization of OZ439.

After intestinal fluid was added, the suspension remained in a water bath at 37 °C, and 0.8 mL aliquots were removed at t = 0, 0.25. 0.5, 1, 3, 6, and 24 h (Step 4, Fig. 3). Aliquots, which contained bile salts, dissolved OZ439, and nanoparticles, were centrifuged in an Eppendorf Centrifuge 5430R at 28,000 rpm for 10 min to pellet nanoparticles (Step 5, Fig. 3). The supernatant was then removed, frozen, and lyophilized (Step 6, Fig. 3). The lyophilized powder was resuspended in a mixture of acetonitrile and THF (90/10, v/v), which dissolved any OZ439 present, but not residual bile salts. This suspension was sonicated to help dissolve OZ439, then centrifuged to pellet the insoluble bile salts (Step 7, Fig. 3). The supernatant was removed and filtered through a GE Healthcare Life Sciences Whatman™ 0.1 µm syringe filter. OZ439 concentration was determined by high performance liquid chromatography (HPLC) using a Gemini C18 column (particle size 5 μm, pore size 110 Å). The OZ439 detection method used an isocratic mobile phase of 99.95%/0.05% acetonitrile/trifluoroacetic acid at 45 °C and a detection wavelength of 221 nm. OZ439 concentration was calculated from a standard curve. Measurements were performed in triplicate.

Figure 3 shows a flow diagram of the in vitro dissolution test conditions and subsequent OZ439 separation train. The loss of OZ439 throughout the steps was minimal; in several instances, an amount of dissolved OZ439 over 98% of the theoretical maximum was observed.

Long-term powder stability

For a nanoparticle formulation in dry powder form to be effective at combatting malaria in the developing world, it must retain its superior drug solubilization properties through long-term storage in hot, humid conditions. The tests described below were intended to rapidly age the powders in harsh conditions before assessing their physical characteristics and dissolution kinetics. A future study in the formulation’s development will include temperature cycling and use commercially suitable storage containers and conditions that reflect the real world conditions. Here, three phases of experiments were employed to assess powder stability. First, vials containing lyophilized OZ439 NPs were placed uncapped in an oven at 50 °C and 75% relative humidity (RH). After 1 day, and again after 1 week, aliquots of powder were removed and their OZ439 dissolution kinetics were measured using the protocol above.

In the second phase, vials of spray dried OZ439 NPs were placed in the same conditions (uncapped, 50 °C, 75% RH). OZ439 dissolution was measured after 1, 3, 7, 14, 21, and 28 days. At each time point, some powder was removed for quantification by XRPD, DSC, and titration to determine water content. This phase is referred to as the ‘28-day time course.’

In the third phase, referred to as the ‘6-month time course,’ spray dried OZ439 NPs in capped vials (hand tight, without sealant or tape) were placed in an oven at 40 °C and 75% RH. After 3, 7, 14, and 28 days, and 2, 3, and 6 months, a vial was removed, and OZ439 solubilization was tested and XRPD was performed. In addition, at t = 0, 2, and 6 months, water content was determined and DSC was performed.