Transformation of Ritonavir Nanocrystal Suspensions into a Redispersible Drug Product via Vacuum Drum Drying

The present study explored vacuum drum drying (VDD) as potential drying technique for the solidification of crystalline ritonavir nanosuspensions prepared by wet-ball milling. In detail, the impact of drying protectants (mannitol, lactose, trehalose) added to the ritonavir nanosuspension was assessed in dependence of the drum temperature with respect to processibility via VDD, resulting intermediate powder properties, remaining nanoparticulate redispersibility and crystallinity. A clear impact of the glass transition temperature (Tg) of the drying protectant on the redispersibility/crystallinity of the VDD intermediate was observed. Increased Tg of the drying protectant was associated with improved redispersibility/crystallinity at a defined drum temperature. Consequently, the high Tg-substance trehalose and lactose showed a better performance than mannitol at higher drum temperatures. However, the processability and related powder properties were not in accordance with this observation. Mannitol containing formulations showed superior processibility to those containing trehalose/lactose. Moreover, the impact of the tableting and encapsulation process on the redispersibility of the VDD intermediate was studied for a selected formulation. Neither process demonstrated a negative impact on redispersibility. In conclusion, vacuum drum drying is a promising drying technique for the solidification of nanosuspensions to result in dried powder still containing ritonavir nanoparticles while demonstrating acceptable to good downstream processibility to tablets/capsules.


INTRODUCTION
The importance of addressing the poor water-solubility of drug candidates in pharmaceutical development has become more pronounced in the last years, since nearly 90% of the drug candidates are poorly soluble, resulting in limited bioavailability (1)(2)(3). One strategy to tackle the solubility issue is the nanocrystal approach, whereby the crystalline drug substance is nanosized to improve dissolution behavior and saturation solubility according to the Noyes-Whitney and Ostwald-Freundlich principles (4)(5)(6). The nanocrystal formulations are generally based on a liquid, aqueous nanocrystal suspension (nanosuspension) in the nanometer size range (100-1000 nm). However, nanoparticles in aqueous media require stabilization since nanoparticles are much less stable than microparticles due to the Gibbs free energy contribution. Two types of stabilizers with different functional principles are described in literature: ionic stabilizers via thermodynamic/electrostatic stabilization, and steric stabilizers via kinetic stabilization (7,8). The combination of both is most commonly utilized, demonstrating the highest stabilization effectiveness due to a synergistic effect which is also referred to as electrosteric stabilization (9)(10)(11).
Nanosuspensions can be prepared by either top-down (e.g., wet ball milling) or bottom-up (e.g., precipitation) approaches (12,13). However, in the pharmaceutical industry, top-down approaches are far more relevant due to their simplicity, reproducibility and scalability (9,13). The most prominent methods are NanoCrystal® (wet ball milling) and IDD-P™ (insoluble drug delivery microparticle technology, high pressure homogenisation) (13) which are used in the manufacture of drug products such as Invega® Sustenna™ (Paliperidone palmitate via NanoCrystal®, Janssen 2009) or Triglide® (Fenofibrate via IDD-P™, Sciele Pharma, Skye Pharma 2005) (9). Nanocrystals in aqueous suspensions are still associated with a certain risk of instabilities, either physical (e.g., Ostwald ripening and agglomeration), chemical (e.g., hydrolysis), or a risk for microbial growth leading to limited product shelf life (14)(15)(16). Another disadvantage of the administration of nanosuspensions as a liquid dosage form is the error-prone dosing step for patients, which may affect patient compliance and trigger the need for dosing devices. These disadvantages can be overcome by transforming the liquid nanocrystal suspension into a solid dosage form such as tablets or powder filled capsules. State-of-the-art solidification (drying) techniques include spray-drying, spray-coating (also termed spray granulation), and freeze drying (11)(12)(13).
In general, drying is a critical and essentially destabilizing procedure for the nanocrystal system which may lead to particle agglomeration and/or aggregation as well as crystal growth followed by sedimentation or flocculation. Consequently, redispersibility of nanoparticles upon reconstitution of the dried powder could be affected, which in turn reduces the beneficial effect of nanosizing on dissolution. For this reason, drying protectants are usually added to the final nanosuspension prior to the drying process to avoid particle growth. Common drying protectants include soluble sugars, such as lactose, sucrose or trehalose, or sugar alcohols like mannitol (14,17). However, no nanocrystalbased drug product was approved by the FDA from 2009 to 2018 although drying protectants enable processability (9). This demonstrates how challenging and less economically efficient the common drying processes for solidification of nanocrystal suspensions are compared to other enabling technologies such as solid dispersion technology for solubility enhancement. This is particularly relevant in light of the observation that nanosizing technology is frequently used for early toxicology studies supply, which would be a straightforward approach for FIH development. Recently, vacuum drum drying (VDD) has been introduced as alternative approach to manufacture amorphous solid dispersions showing benefits especially compared to spray drying (18,19). This technology is well-established in the food industry, but rarely known in the development of pharmaceuticals (20). The manufacture of ASDs as one enabling formulation principle consists of the embedment of the drug substance molecularly dispersed (i.e., amorphous) in a matrix polymer. Using vacuum drum drying as technology requires the dissolution of the drug substance and respective excipients in a common organic solvent.
In contrast, the nanocrystal approach essentially bases on the size reduction of a crystalline drug substance into nanosized particles (nanocrystals) in aqueous media followed by a solidification (drying) step. Consequently, the focus of the present case study was to investigate the suitability of vacuum drum drying as solidification (drying) technique for nanosuspensions with the potential to overcome obstacles and disadvantages of currently available drying technologies by, e.g., not needing a secondary drying step and by showing no viscosity related limitations for the solution to be dried and potentially higher yields. Thus, the suitability of vacuum drum drying for another enabling formulation principle, entirely different to the ASD approach, was examined to broaden the applicability in the pharmaceutical development.
The nanosuspension analyzed in this study consisted of ritonavir as model drug substance, sodium dodecyl sulfate as ionic stabilizer, and copovidone as steric stabilizer. More detailed, the impact of the drying protectant (mannitol, lactose, trehalose) on the processibility via VDD, on the dried powder properties, the remaining nanoparticulate redispersibility and crystallinity was assessed in dependence of the drum temperature. Subsequently, the most promising formulation was selected for further downstream processibility evaluation, with a particular focus on redispersibility. Therefore, the impact of compaction pressures during tableting was investigated as well as the impact of the encapsulation process on a pilot scale machine.

Preparation of Ritonavir Nanosuspensions by Wet Ball Milling
The ritonavir nanosuspension was prepared by wet ball milling (top-down approach, batch sizes 1.50-1.85 kg). Ritonavir (15% w/w) and 0.5 mm zirconium oxide beads as grinding media (bead to ritonavir ratio: 1:18) were added to a stabilizercontaining solution (SDS (1% w/w) and copovidone (3% w/w)) into a 5-L HDPE (high-density polyethylene) bottle. The nanosizing was performed using a tumble blender (Turbula blender T10B, Willy A. Bachofen AG Maschienenfabrik, Muttenz, Switzerland) at 45 rpm for 69 h. Then, the zirconium beads were separated via filtration using a sieve with 200 μm mesh size. Prior to vacuum drum drying, drying protectants (mannitol, lactose, or trehalose; amount: 7.5%, 10%, 15%, or 25% w/w (see Table I)) as well as copovidone (7% w/w) were added to the ritonavir nanosuspension while stirring on a magnetic stirrer (IKA GmbH & Co KG, Staufen, Germany). Copovidone was added to increase the viscosity of the liquid formulation (drying dispersion) and thus, to increase the adhesion of the solution to the drums.
The nanosuspension formulation and the liquid formulations (drying dispersions) including drying protectant and copovidone were selected based on prior knowledge and formulation screening data (data not shown). The drying protectants investigated in the present study are commonly used excipients in the solidification of crystalline nanosuspensions (see "Introduction").

Characterization of Nanosuspensions
Particle Size Analysis by Dynamic Light Scattering Dynamic light scattering (DLS) method was applied using a Zetasizer Ultra (Malvern Instruments GmbH, Herrenberg, Germany) to determine the z-average and the polydispersity index (PDI) of the nanosuspensions after wet-ball milling.
z-average represents the hydrodynamic diameter, and PDI expresses the width of the particle size distribution. The samples were diluted in water (1:20), and polystyrene single-use cuvettes (DTS0012) were used. The measurements were performed in back scatter mode (173°) as triplicates at 25°C prior to an equilibration time of 120 s. The results were analyzed using ZS Xplorer software (version 1.3.2.27).
Particle Size Analysis by Laser Diffraction Laser diffraction particle size analyzer (Mastersizer 3000, Malvern Instruments GmbH, Herrenberg, Germany) equipped with the automated dispersion unit "HydroMV" module was used to determine the particle size distribution of the nanosuspensions. For the measurements, nanosuspension was added to water until a laser obscuration of approximately 2-2.5% was reached. Data were analyzed according to the MIE theory using the Mastersizer 3000 Software (version 3.71). Measurements were performed as triplicates and averaged.

Redispersibility of VDD Intermediates/Tablets/Capsules
For redispersibility evaluation, the VDD intermediates, tablets, or capsule-fillings were dispersed in an appropriate amount of water targeting the ritonavir concentration of the original nanosuspension (ritonavir: 15% w/w). The resulting suspension was mixed using a vortexer (IKA Vortexer VG3, Staufen, Germany) and subsequently characterized by laser diffraction and/or dynamic light scattering (see "Characterization of Nanosuspensions").

Redispersibility by Particle Fractions in Submicron Range
The redispersibility was evaluated by means of laser diffraction to increase the understanding of the agglomeration state of the ritonavir particles, since larger particles could sediment during DLS analysis not being detected then. The percentage of particles below 1 μm in a cumulative volume-based particle size distribution was selected as criterion to determine the redispersibility.

Redispersibility Index
For better comparability of the redispersed suspension with the initial nanosuspension, the redispersibility index (RDI) was calculated by normalizing the particle size describing variable (z-average or d 50 ) to the respective variable of the initial nanosuspension.
Consequently, an RDI value close to 1 indicates a sufficient preservation of the nanoparticulate drug substance particles after solidification. For processability evaluation, the particle size describing variable of the tablets/capsule powder fill was compared with the respective one of the VDD intermediate (powder).

Characterization of VDD Intermediates
Bulk/Tapped Density Determination of bulk and tapped density was performed using tapped density tester (Pharmatest Apparatebau AG, Hamburg, Germany) according to Ph. Eur. 2.9.34 (method 1). VDD intermediate was filled into a 250ml graduated cylinder and the mass and bulk/tapped volume occupied by the material was determined. All measurements were conducted as triplicates.
Particle Size Distribution Determination of particle size distribution of VDD intermediates (powder) was performed using a laser diffraction particle size analyzer (Mastersizer 3000, Malvern Instruments GmbH, Herrenberg, Germany) equipped with a dry powder disperser module Aero S. The samples (approximately 2-5°g) were dispersed with 0 bar pressure and measured as triplicates. Data were analyzed using the Mastersizer 3000 Software (version 3.71) according to the Fraunhofer approximation.
Loss on Drying Determination of moisture content via loss on drying (LOD) method was performed using a halogen moisture analyzer (HB43-SSD, Mettler-Toledo GmbH, Giessen, Germany). The samples (approximately 5.5-6.1 g) were heated to 105°C and held until mass was constant within ± 1 mg for 100 s. The VDD intermediates were measured as triplicates.
(1) RDI = d 50 (or z − average) redispersed d 50 (or z − average) initial Crystallinity and Glass Transition Temperature by Differential Scanning Calorimetry Quantification of ritonavir-related crystallinity as well as determination of the glass transition temperature (T g,wet ) was performed via differential scanning calorimetry (DSC) using a Mettler-Toledo DSC 3+ (Mettler Toledo, Gießen, Germany) equipped with an auto-sampler. All DSC samples (VDD intermediates) were scanned at 1.  Figure OR1.
To verify the DSC results on quantification of the crystalline ritonavir content, powder X-ray diffraction (PXRD) was performed for one selected VDD intermediate (Man7.5/ Tre7.5_75, see supplemental data).

Short-Term Stability Focusing on Redispersibility
The physical stability of the selected VDD intermediate for downstream evaluation was determined by DLS and LD after storage for 2 and 6 months at uncontrolled conditions at room temperature and relative humidity (approximately 45-50%) and compared with initial nanosuspension data and initial powder characterization after solidification (drying).

Downstream Processability
Tabletability and Tableting For tabletability evaluation, round, biplane tablets (10 mm, mass: 200 mg, n = 6) of selected formulations (small-scale runs; see Table I)  Quali-V capsules in opaque grey targeting a fill weight of 157.2 mg (corresponding to 50 mg ritonavir). Encapsulation process was performed using an automated capsule filling machine (Modu-C LS, Harro Höfliger, Allmersbach im Tal, Germany) equipped with an inline in-process (weight) control unit. Process parameters were set as follows: 20 cycles/ min, 100% check weighing (net weight). In-process control samples were taken at start, middle, and end of the encapsulation process and evaluated according to Ph. Eur. 2.9.5 (uniformity of mass of single-dose preparation). In addition, a process sample (n = 100) was taken and analyzed (gross weight). The machine protocol was used for process assessment (net weight).

Characterization of Ritonavir Nanosuspensions
Wet ball milling of ritonavir was successfully conducted in 5 sub-batches (runs) at different batch sizes (1.5-2.0 kg). The particle size analysis results are shown in Figure 1. The z-average values determined via dynamic light scattering (DLS) were below 400 nm (in a range of 300-370 nm), and the PDIs were below 0.15 indicating monodispersed ritonavir nanosuspensions (see Figure 1a). Laser diffraction (LD) analysis confirmed the DLS results showing the absence of large particles (e.g., agglomerates), which could be potentially missed via DLS due to sedimentation. The d 50 values obtained by LD were in a range between 113 and 315 nm and more than 98.5% of the particles were in the submicron range (<1 μm) (see Figure 1b).

Impact of Drying Protectant and Drum Temperature on Processability
A clear dependency of the drying protectant used on processability could be observed as shown in Table II. Yield values were consistently low for lactose-containing formulations ranging from 53 to 62%. Mannitol-containing formulations showed yield values of 65-86% indicating a better process performance compared to lactose-containing formulations, whereas trehalose-containing formulations showed a pronounced drum temperature dependence: the higher the drum temperature, the better the yield (75°C: 63.3%, 105°C: 92.8%). Combining mannitol with either lactose or trehalose as drying protectants resulted in formulations with acceptable yield values of 65-87% for small-scale runs on a pilot scale VDD. The LOD values were between 1 and 3% for all other formulations except for the trehalose formulation (processed at 75°C with LOD of 4.90%) and the mannitol formulation (processed at 55°C with LOD of 4.87%).

Impact of Drying Protectant and Drum Temperature on Powder Properties of Dried Intermediates
Dried VDD intermediates have been characterized with respect to flowability (ring shear analysis), powder density (bulk/tapped), and solid particle size distribution (PSD). The results are summarized in Table II. The results indicated an impact of drying protectant type on flowability. All mannitol containing formulations exhibited easy flowing properties, whereas the lactose or trehalose containing powders showed cohesive flow independent of the drum temperature. Flow function coefficient values (FFC) for trehalose containing formulations indicated cohesive flow at lower temperatures and borderline easy flow at the highest drum temperature.
The bulk density values (Table II) were most favorable for further downstream processing for mannitol containing formulations with values at around 0.30-0.46 g/cm 3 . The lactose and trehalose containing formulations showed lower bulk density ranging from 0.11 to 0.17 g/cm 3 indicating a fluffy powder. The formulations with two drying protectants, either mannitol/lactose (Man7.5/Lac7.5_75: 0.35 g/ cm 3 ) or mannitol/trehalose (Man7.5/Tre7.5_75: 0.35 g/ cm 3 ), resulted in powders with bulk density values in the same ranges of the pure mannitol containing formulation (Man15_75: 0.34 g/cm 3 ). Consequently, mannitol might be the dominant component within the formulation with respect to bulk density. This could be confirmed by the results of mannitol containing formulations at different mannitol levels: the higher the mannitol content within the dried product, the higher the bulk density: The bulk density for Man10_75 was 0.30 g/cm 3 , that for Man15_75 was 0.34 g/cm 3 , and the one for Man25_75 was 0.46 g/cm 3 . For mannitol and lactose containing formulations, a minor dependence between drum temperature and bulk density values could be observed: the lower the drum temperature, the higher the bulk density. For trehalose, no trend could be observed.
The particle size distribution of the intermediates indicated relatively large particles of d 90 values even above 1000 μm due to the selected sieve. However, trehalose and lactose containing intermediates tended to lower d 90 values at around 600-650 μm for trehalose and 535-844 μm for lactose compared to mannitol (981-1120 μm). In addition, the mannitol level within the formulation impacted the d 90 value: the higher the mannitol content, the higher the d 90 value (Man10_75: 796 μm; Man15_75: 981 μm, Man25_75: 1330 μm). Furthermore, mannitol seemed to substantially impact the particle size distribution when combined with other drying protectants.

Impact of Drying Protectant and Drum Temperature on Redispersibility
The impact of the drying protectant (mannitol, lactose, trehalose) at a defined amount (15% w/w within the liquid formulation (drying dispersion)) was assessed with respect to particle size and PDI after redispersing the VDD intermediate (redispersibility) dried at various drum temperatures via LD. Process conditions and the results are summarized in Table II. A clear trend could be observed for all three drying protectants investigated: with increasing drum temperature, the number of particles in the submicron range decreased, and the RDI increased indicating particle agglomeration/   Figure 2 illustrates the T crit and T g,wet for the respective formulations and the T g,dry of the pure drying protectants according to literature (12). A clear dependency between the T g,dry values of the respective drying protectants, the resulting formulation T g,wet values, and the T crit values could be observed. Formulation Man15 with the lowest T g,wet showed the lowest T crit (55°C) meaning that a redispersible intermediate is feasible at the lowest drum temperature tested. Consequently, applying drum temperatures above the T crit during solidification would lead to a dried, less to non-redispersible product. In contrast, dried material with trehalose displayed the highest T g,wet value, and also the highest T crit value with 90°C. However, the identified T crit values were approximately 30-35°C above the wet T g,wet for all formulations. Figure OR1 (Online Resource) displays the DSC thermograms of the pure microcrystalline ritonavir as reference for quantification of the remaining ritonavir-related crystallinity fraction (n = 2), and of the vacuum drum dried intermediate (Man7.5/Tre7.5_75_large) representative for all DSC measurements (data not shown). The crystalline fraction within the Man7.5/Tre7.5_75_large VDD intermediate was 23.2% corresponding to 73.0% remaining crystallinity. This is in accordance with the small-scale batch data (75.1%) of the same formulation. Additionally, an estimation of ritonavir-related crystallinity was determined via powder X-ray diffraction (PXRD) for the formulation Man7.5/ Tre7.5_75_large. The remaining ritonavir-related crystallinity value was 89.93 % (see Figure OR2 (Online Resource)). Figure 3 shows the impact of the drum temperature during vacuum drum drying on ritonavir-related crystallinity determined via DSC for formulations containing mannitol, lactose, or trehalose as drying protectant. Data indicate a dependency between drum temperature and ritonavir-related crystallinity: the higher the drum temperature for the respective formulation, the lower the crystalline ritonavir content within the VDD intermediate. A strong decrease in crystallinity could be observed for VDD intermediates processed at drum temperatures above T crit . The crystallinity values at T crit were in a range of 75-85%.

Short-Term Stability of Selected Formulation
The short-term stability study was carried out on the selected formulation Man7.5/Tre7.5_75 (see "Downstream Processability of Selected Formulation"). The VDD intermediate was analyzed after 2-and 6-month storage at uncontrolled conditions (room temperature; relative humidity of approximately 45-50%) using LD and/or DLS. Results were compared with VDD intermediate at study start (T0) and the corresponding nanosuspension. Results are given in Table III comprising the d 50 values, the cumulative number of particles in the submicron range, and the z-average and PDI values. The d 50 values ranged from 111 to 139 nm indicating no distinctive change over time regarding particle size. The number of particles in the submicron range were consistently above 97% and comparable to the corresponding nanosuspension with an initial value of 98.2%. However, z-average values indicated a slight shift to larger particles during storage: 299 nm (nanosuspension), 330 nm (after 2 months), and 353 nm Fig. 2 Critical process (drum) temperature (T crit ) and glass transition temperature (T g,wet ) of formulations containing different drying protectants (mannitol, lactose, trehalose) and T g,dry of the pure drying protectants according to literature (12) Fig. 3 Ritonavir-related crystallinity dependent on drum temperature of formulations containing mannitol (Man15), lactose (Lac15), and trehalose (Tre15) as drying protectant. Critical drying temperature (T crit ) for each formulation displayed in red (after 6 months). Consequently, it is recommended to store the dried product at lower temperatures and low humidity to avoid further particle agglomeration and/or crystal growth.

Downstream Processability of Selected Formulation
Formulation Man7.5/Tre7.5_75 was selected as prototype formulation to evaluate downstream processability comprising several benefits compared to other formulations tested. It showed good remaining nanoparticulate redispersibility, remaining crystallinity, and favorable powder properties such as bulk density (see Table II). The small-scale batch intermediate was used for tabletability evaluation to ensure proper comparability to the other formulations processed into powder intermediates using similar equipment (manually milled VDD intermediate). The large-scale batch intermediate was used for encapsulation experiments. Figure 4a shows the tabletability plot (tensile strength vs compaction pressure) of selected formulations simulating the rotary press Korsch XL100 at 20 rpm turret speed (linear speed: 124 mm/s). All formulations were easily compressible leading to tablets with sufficient hardness even at low compaction pressures. Still, differences were observed depending on the drying protectant used in the formulation composition. Trehalose showed the best tabletability followed by lactose. Mannitol exhibited the least favored tabletability profile, but still showed sufficient tensile strength. Mannitol-containing formulations with lactose or trehalose in combination revealed comparable tabletability to the mannitol-only formulation. Consequently, mannitol affected tabletability most. Moreover, no tablet defects were observed for all formulations tested. Figure 4b shows the impact of tensile strength on tablet disintegration for different formulations. All tablets showed fast disintegration time (< 12.5 min), which decreased with decreasing tensile strength.

Impact of Tableting Process on Redispersibility
Tablets of different formulations with a tensile strength of 1.2 MPa were investigated regarding redispersibility after tableting by laser diffraction. The results are summarized in Table IV. The d 50 values of the redispersed tablets were similar to the initial VDD powder intermediates. RDI values of 1.0-1.1 indicated no substantial change in PSD of redispersed particles with a large number of particles (> 93%)  . 4 a Tabletability plots of selected formulations simulating a Korsch XL100 at 20 rpm (n = 6); b Disintegration time of selected formulations compressed to tablets of defined tensile strengths (n = 6); Man = mannitol, Lac = lactose, Tre = trehalose smaller than 1 μm. Consequently, no impact on tableting on redispersibility could be observed. In addition, the impact of tableting on redispersibility was studied with tablets of different tensile strengths via DLS and LD for a selected formulation (Man7.5/Tre7.5_75). As shown in Figure 5, the z-average values by DLS were in a comparable range (342-367 nm) for all tensile strengths. Measured PDIs (0.100-0.175) indicated monodispersed nanosuspensions. RDI values of 1.0 to 1.1 demonstrated good redispersibility. LD results were in accordance with those from DLS measurements: the d 50 values varied from 196 to 239 nm with 95.1-99.8% of the particles in the submicron range. The RDI values of LD measurement did not indicate any particle size change induced by the tableting process.

Impact of Encapsulation Process on Redispersibility
The selected VDD intermediate (Man7.5/Tre7.5_75) was successfully encapsulated into size 0 HPMC capsules targeting a fill weight of 157 mg (RTV dose of 50 mg). Approximately 1000 capsules were manufactured. The results of the analyzed in-process control and process samples combined with machine protocol data are summarized in Table OR2 (Online Resource) reflecting an acceptable encapsulation process.
Redispersibility of the capsule powder fill (powder) was assessed via DLS and LD for the in-process control samples (see Figure 6). The z-average values ranged from 320 to 333 nm and the PDI from 0.114 to 0.126 assuming monodispersed nanosuspensions. The RDI values were constantly at 1.0. The particle size analysis data via LD were in accordance with the DLS data: the d 50 values were between 135 and 140 nm with 99% of the particles in the submicron range without variation with respect to processing time. And the RDI values were at 1.0 indicating no change in particle size compared to the VDD powder. Thus, the encapsulation process did not impact the remaining nanoparticulate redispersibility of the capsule powder fill.

DISCUSSION
Ritonavir nanocrystal suspensions could be successfully manufactured with good reproducibility via wet-ball milling using zirconium oxide beads using a classic tumble blender.
The data revealed that processibility of the tested nanocrystals containing liquid formulations (drying dispersions) during vacuum drum drying was impacted by the drying protectant. Mannitol was identified as best drying protectant for product solidification in terms of visual behavior on the drums, LOD, and yield. Similar observations were made by Chaubal, Popescu (12) comparing lactose, mannitol, sucrose, and dextrose containing spray dried powders of itraconazole nanosuspensions: mannitol was rated as most favorable carrier for spray drying of nanoparticles providing most desirable particle morphology, flowability, and LOD values.
The redispersibility was mostly affected by the interplay between drying protectant and drum temperature. An individual critical process temperature (T crit ) correlating with the  (14). Researchers reported that the outlet temperature during spray drying seemed to be critical and correlated with the T g,wet of the formulation, which was shown to be drug load related. The temperature difference between T g,wet of the formulation and T crit (outlet) was similar for all formulations tested (approximately 20-25°C). In the present study, the difference between T g,wet of the formulation and T crit (drum) was 30-35°C and thus, approximately 10° higher compared to the spray drying data. This might be explained by the fact that the drum temperature did not reflect the product temperature, which is probably lower due to the cooling effect during water evaporation. In contrast, the outlet temperature during spray drying is much better linked to the real product temperature. Consequently, it can be assumed that a T crit related to the T g,wet of the formulation might be process independent. An explanation for the temperature impact on redispersibility could be as follows: process temperatures above the T g,wet of the material knowingly increase the fluidity of the material and potentially facilitate nanocrystals aggregation and growing. Interestingly, Malamatari et al. (22) identified the ratio of drying protectant to drug as another important formulation-related factor for redispersibility. The present study confirmed their findings, as the amount of submicron particles was increased at higher mannitol to ritonavir ratio (20% for Man15_75, 40% for Man25_75) at constant drum temperature. This stabilizing effect was likely caused by steric hindrance by the drying protectant. Zuo et al. explains this steric hindrance as follows: the presence of water-soluble additives such as Mannitol could form hydrophilic excipient bridges interconnecting the nanoparticles and thus, avoiding crystal-to-crystal contact and in the end crystal growth. The same stabilizing effect is seen in lyophilization processes where mannitol is used as so called lyo-protectant (23). Surprisingly, Man10_75 showed better redispersibility compared to Man15_75, despite containing a lower amount of drying protectant. In this case, the T g impact would be more pronounced as stabilizing principle than the steric hindrance. Reducing the mannitol content resulted in a higher relative content of copovidone in the formulation, which has a higher T g,dry (101°C) compared to mannitol. Consequently, the T g,wet of the formulation might be higher and thus, leading to a higher stabilizing effect against temperature. In addition, copovidone acts as matrix polymer still providing sufficient stabilization. Hence, the stabilizing effect of drying protectants on the nanocrystals might be dominated by two principles: steric hindrance (acting as spacers by building excipient bridges) and T g influence. However, other components of the formulation might influence the T g significantly as well (e.g., drug substance, ionic or polymeric stabilizers or adhesion enhancers). The remaining ritonavir-related crystallinity data determined via DSC showed a clear ritonavir melting peak visible in the thermograms of all analyzed samples. However, a temperature shift to a lower melting temperature was observed for the VDD intermediates containing ritonavir nanoparticles compared to the pure microcrystalline ritonavir measurements. Based on literature, melting point depression is expected with reduction of crystal particle size as described by the Gibbs-Thomson equation (24). In fact, this has been already reported for several nanosuspensions dried via spray drying in previously published studies (25,26). Consequently, measuring the remaining crystallinity at nanometer size via DSC might potentially not reflect the true crystallinity. Few percent of crystallinity might get lost due to fast melting of nanocrystals at even lower temperatures which is also implied by the left skewed melting peak in the VDD intermediate thermograms. In fact, even PXRD data represent only an estimation, because a reference standard had to be used for quantification. Consequently, the true However, DSC data clearly indicated that a portion of the ritonavir was converted into a non-crystalline (most likely amorphous) form even at the lowest process (drum) temperatures. This observation was in accordance with the results of a recently published study, where a risperidone nanosuspension was processed by spray drying (27). Kayaert, Van den Mooter (28) stated that the cause of amorphization is most likely the interplay between drug and stabilizer during drying, rather than the nanosizing via wet-ball milling. Indeed, if the drug substance is soluble in the stabilizer, especially if the stabilizer is a polymer, it enhances the probability of an amorphous layer formed at the interface. In the present formulations, copovidone was used to increase the adhesion to the drums of the vacuum drum dryer, next to its function as polymeric stabilizer of the liquid nanosuspension. Consequently, the level of copovidone within the final dried intermediate was quite high (25% w/w) enabling the solubilization process of ritonavir. This solubilization was even more enhanced by the presence of nanocrystals instead of microcrystals. In addition, this study clearly showed a drum temperature dependency for the remaining crystallinity: the higher the drum temperature, the higher the amorph fraction. Interestingly, the critical drum temperature for remaining crystallinity corresponded with the T crit value for redispersibility.
Since non-crystalline API, e.g., amorphous API, can recrystallize during storage (28), this partial change in solid state might have an impact on stability. Consequently, different solid states after manufacturing and after drying of the nanosuspension should be avoided to ensure stability as key design requirement. This might be even more critical for substances showing pronounced polymorphisms. Thus, future research should investigate the impact of non-crystalline API within a solidified nanocrystal drug product on storage stability.
The short-term stability evaluation of the present study indicated a slight increase in particle size determined via DLS (see Table III). However, an increase in particle size could not be detected via LD. Kumar et al. (29) found out that all indomethacin nano-formulations were stable after spray drying and during storage stability which contained small molecular weight sugars such as mannitol, lactose, and trehalose. However, the ratio between the drying protectant and API was much higher (1 to 5 w/w) compared to the formulation tested in this present study (1 to 1.3 w/w). A higher drying protectant to API ratio is known to be beneficial ensuring better steric stabilization, and thus, redispersibility as stated by Malamatari et al. (22). Yet, the API load would be significantly lower.
As mitigation concept for crystal growth/aggregation, stability, and amorphization, the T g of the formulation should be determined prior to the drying step to choose the processing temperature, accordingly, meaning below the T crit of the respective formulation. This applies presumably for both, spray drying and vacuum drum drying. However, the composition of the nanosuspension prior drying needs to be selected carefully considering the type of drying protectant, its T g , and the ratio of drying protectant to API. Drying protectants with higher T g might be preferred enabling lower drying protectant to API ratio, and thus, higher API loads of the final dosage form. Moreover, the solubility of the API within the polymeric stabilizer needs to be assessed to avoid unintended amorphization of the API.
The compression into tablets revealed a dependence on the drying protectant in the formulation. This might be related to the compression behavior of the neat drying protectant and the particle size and shape of the VDD intermediates. Nevertheless, all formulations showed acceptable tabletability, since resulting tablets exhibited sufficient tensile strength even at low compaction pressures.
Results from the redispersibility assessment indicated that tableting did not impact redispersibility for all formulations at a tensile strength of 1.2 MPa. This was surprising, since nanosized ritonavir particles were assumed to proximate to each other during the tableting process. But even at higher compaction pressures and higher tablet tensile strength respectively, only disintegration was affected, however, not redispersibility. Moreover, it could be demonstrated that encapsulation of VDD intermediate to capsules did not affect redispersibility of the final drug product.
To sum up, the drying process of a stabilized nanosuspension seems to be the most critical step during the manufacture of a nanocrystal drug product. The selection of the formulation components is important for several reasons. Selected excipients should enable the following: -Stabilizing the liquid nanosuspension by preventing particle growth, agglomeration, and precipitation -Enabling the spreading of the nanosuspension, and in turn ensure uniform drying on the heated drums -Achieving acceptable flow and density of the dried and subsequently screened intermediate to ensure downstream processability for encapsulation or compression.

CONCLUSION
The present study demonstrated the feasibility of using vacuum drum drying (VDD) for the solidification of ritonavir nanosuspensions resulting in redispersible solids. VDD offers advantages for downstream processing. First, no second drying step is required. Second, the powder properties of the dried intermediate, such as particle size distribution, can be adjusted during milling and by this, potentially optimized for subsequent encapsulation or compression into tablets, which was demonstrated for a selected model formulation. Moreover, first insights were gathered on the interplay between formulation composition and VDD process conditions (drum temperature), on the resulting impact on powder redispersibility, and on the remaining ritonavir related crystallinity as follows: • For all studied formulations redispersibility and DScrystallinity substantially decreased exceeding a formulation specific drum temperature (T crit ). • As T crit is formulation dependent, it should be identified for each formulation as part of the process development. • T crit might be correlated with the glass transition temperature (T g,wet ) of the formulation, which is mostly dominated by the T g of the pure drying protectant. However, this needs further mechanistic clarification. • Particle growth during drying can be prevented by taking advantage of the principles of steric hindrance in combination with possibly high T g of the formulation leading to less fluidity. • As the stabilizers for the nanosuspension will affect the overall T g , the type and amount should be carefully selected for formulation composition.