Development of Remdesivir as a Dry Powder for Inhalation by Thin Film Freezing

Remdesivir exhibits in vitro activity against SARS-CoV-2 and was granted approval for emergency use. To maximize delivery to the lungs, we formulated remdesivir as a dry powder for inhalation using thin film freezing (TFF). TFF produces brittle matrix nanostructured aggregates that are sheared into respirable low-density microparticles upon aerosolization from a passive dry powder inhaler. In vitro aerodynamic testing demonstrated that drug loading and excipient type affected the aerosol performance of remdesivir. Remdesivir combined with optimal excipients exhibited desirable aerosol performance (up to 93.0% FPF< 5 µm; 0.82 µm mass median aerodynamic diameter). Remdesivir was amorphous after the TFF process, which benefitted drug dissolution in simulated lung fluid. TFF remdesivir formulations are stable after one month of storage at 25 °C/60% relative humidity. An in vivo pharmacokinetic evaluation showed that TFF remdesivir–leucine was poorly absorbed into systemic circulation while TFF remdesivir-Captisol® demonstrated increased systemic uptake compared to leucine. Remdesivir was hydrolyzed to the nucleoside analog GS-441524 in the lung, and levels of GS-441524 were greater in the lung with leucine formulation compared to Captisol®. In conclusion, TFF technology produces high-potency remdesivir dry powder formulations for inhalation that are suitable to treat patients with COVID-19 on an outpatient basis and earlier in the disease course where effective antiviral therapy can reduce related morbidity and mortality.


Introduction
The coronavirus disease 2019 (COVID- 19) is an ongoing worldwide pandemic. As of September 2020, laboratory-confirmed cases have been reported in 213 countries and territories with more than 30 million reported cases and close to one million reported deaths [1]. Although this disease is asymptomatic to mild in most people, in some cases, it can develop into pneumonia, acute respiratory distress syndrome (ARDS) and multi-organ dysfunction [2]. While effective treatments for COVID-19 are urgently needed, including therapeutics and vaccines, currently no drugs are approved to treat COVID-19. Only one drug, remdesivir that is administered by intravenous (IV) injection, has been granted emergency use authorization in adult and pediatric patients hospitalized with severe disease. Therefore, additional therapeutics and routes of administration are needed to treat this respiratory virus.

Drug Quantification (HPLC)
The content of remdesivir was determined through analysis with a Thermo Scientific™ Dionex™ UltiMate™ 3000 HPLC system (Thermo Scientific, Sunnyvale, CA, USA) at a wavelength of 246 nm. A Waters Xbridge C18 column (4.6 × 150 mm, 3.5 µm) (Milford, MA, USA) was used at 30 • C and a flow rate of 0.8 mL/min. The isocratic method was performed for 4.5 min using a mobile phase of 50:50 (% v/v) water-acetonitrile containing 0.05% (v/v) TFA. The retention time of remdesivir was approximately~3.5 min. Dimethyl sulfloxide: acetonitrile: water (10:60:30, v/v) was used as diluent. All analyses exhibited linearity in the range tested of 0.2-250 µg/mL. All chromatography data were processed by Chromeleon Version 7.2.10 software (Thermo Fisher Scientific, Waltham, MA, USA).
For quantity and quality analysis of stability samples of TFF remdesivir powders, the Agilent 1220 Infinity II LC system (Agilent Technologies, Inc., Santa Clara, CA, USA) equipped with an Agilent 1290 Infinity II evaporative light scattering detector (ELSD) (Agilent Technologies, Inc., Santa Clara, CA, USA) was utilized. Waters XBridge BEH C18 column (4.6 × 150 mm, 3.5 µm) (Waters Corporation, Milford, MA, USA) was used at 30 • C and a flow rate of 0.8 mL/min. The gradient method with a mobile phase ratio of 5 to 95% acetonitrile in water with 0.1% trifluoroacetic acid was used for total running time of 20 min. In total, 10 µL of each sample was injected. The stability samples were monitored with 246 nm UV and ELSD to detect possible degradants. For ELSD, evaporator and nebulizer temperatures were set at 60 • C, and nitrogen gas flow was 1.6 L/min. The chromatography data were processed by Chemstation version C.01.10 (Agilent Technologies, Inc., Santa Clara, CA, USA).

In Vitro Aerosol Performance
The aerodynamic properties of the TFF powder samples were determined using a Next Generation Pharmaceutical Impactor (NGI) (MSP Corp, Shoreview, MN, USA) connected to a High-Capacity Pump (model HCP5, Copley Scientific, Nottingham, UK) and a Critical Flow Controller (model TPK 2000, Copley Scientific, Nottingham, UK). The pre-separator was not used in this study. To avoid particle bounce, the NGI collection plates were coated with 1.5% w/v polysorbate 20 in 100% methanol and then let the plates dry before analysis. A high-resistance Plastiape ® RS00 inhaler (Plastiape S.p.A, Osnago, Italy) containing size #3 hydroxypropyl methylcellulose capsules (V-Caps ® Plus, Lonza, Morristown, NJ, USA) and 1-3 mg of powder was attached to a USP induction port by a molded silicon adapter, and the powder was dispersed to the NGI at the flow rate of 60 L/min for 4 s per each actuation, providing a 4 kPa pressure drop across the device. The deposited powders from the capsule, inhaler, adapter, induction port, stages 1-7, and the micro-orifice collector (MOC) were collected by diluting Pharmaceutics 2020, 12, 1002 6 of 28 with a mixture of dimethyl sulfoxide (DMSO)/water/acetonitrile (10:30:60 v/v/v). The remdesivir content in the deposited powders was determined using the HPLC method, as described in Section 2.3.
Copley Inhaler Testing Data Analysis Software (CITDAS) Version 3.10 (Copley Scientific, Nottingham, UK) was used to calculate the mass median aerodynamic diameter (MMAD), the geometric standard deviation (GSD), the fine particle fraction (FPF), and the emitted fraction (EF). The EF was calculated as the total amount of remdesivir emitted from the device as a percentage of total amount of remdesivir collected. The FPF of recovered dose was calculated as the total amount of remdesivir collected with an aerodynamic diameter below 5 µm or 3 µm as a percentage of the total amount of remdesivir collected. The FPF of delivered dose was calculated as the total amount of remdesivir collected with an aerodynamic diameter below 5 µm or 3 µm as a percentage of the total amount of remdesivir deposited on the adapter, the induction port, stages 1-7 and MOC.

X-ray Powder Diffraction (XRPD)
The crystallinity of the powders was evaluated using a benchtop X-ray diffraction instrument (Rigaku Miniflex 600 II, Woodlands, TX, USA) equipped with primary monochromated radiation (Cu K radiation source, λ = 1.54056 Å). The instrument was operated at an accelerating voltage of 40 kV at 15 mA. Samples were loaded in the sample holder and scanned in continuous mode with a step size of 0.02 • over a 2θ range of 5-40 • at a scan speed of 2 • /min, and a dwell time of 2 s.

Modulated Differential Scanning Calorimetry (mDSC)
Thermal analysis of powder samples was conducted using a differential scanning calorimeter Model Q20 (TA Instruments Inc., New Castle, DE, USA) equipped with a refrigerated cooling system (RCS40, TA Instruments Inc., New Castle, DE, USA). Samples of 2-3 mg were weighed and loaded into a T-zero pan. The pan with a T-zero hermetic lid were crimped, and a hole was drilled in the lid before placing the pan in the sample holder. To evaluate the glass transition and glass-forming ability type of remdesivir unprocessed powder, samples were heat at a heating ramp rate of 10 • C/min from 25 • C to 150 • C, then cooled down to −40 • C, and then heated again to 250 • C. To investigate the crystallinity of the TFF formulations, samples were heated from 25 • C to 350 • C with a heating ramp rate of 5 • C/min. The scans were performed with a modulation period of 60 s and a modulated amplitude of 1 • C. Dry nitrogen gas at a flow rate of 50 mL/min was used to purge the DSC cell throughout the analysis. Data were processed by TA Instruments Trios V.5.1.1.46572 software (TA Instruments, Inc., New Castle, DE, USA).

Scanning Electron Microscopy (SEM)
Scanning electron microscopy (Zeiss Supra 40 C SEM, Carl Zeiss, Heidenheim an der Brenz, Germany) was used to identify surface particle morphology of the TFF remdesivir powder formulations. A small amount from each of the powder formulations was placed on a carbon tape. A sputter coater was used to coat all samples with 15 nm of 60/40 Pd/Pt before capturing the images.

Nuclear Magnetic Resonance (NMR)
1 H-NMR spectra were obtained using a Varian ® NMR 400 MHz Spectrometer (Agilent Inc., Palo Alto, CA, USA) at 25 • C and were used to confirm the purity of remdesivir in the TFF powders, and to study interactions between excipients and remdesivir of TFF powders. Formulations F7, F8, F9, F10, F14 and unprocessed L-leucine and remdesivir drug substance were dissolved in dimethyl sulfoxide-d6 (DMSO-d6). Solutions were then transferred to 5 mm NMR tubes for NMR data acquisition. Chemical shifts were referenced to a residual solvent, DMSO, at 2.50 ppm. The NMR spectrum peaks of remdesivir in the formulations produced by TFF were compared to those of remdesivir unprocessed powder as received from the vendor. Excipients peaks in TFF remdesivir powders were also compared to those of unprocessed excipients.

Preparation of Simulated Lung Fluid (SLF)
The composition and preparation of simulated lung fluid (SLF) were adapted from Hassoun's study [32]. SLF contained dipalmitoylphosphotidylcholine (DPPC), cholesterol, albumin, transferrin, ascorbic acid and Hank's balanced salt solution (HBSS). The SLF was prepared by two steps including a preparation of liposomal dispersion and freeze drying as described in Hassoun et al. [32]. In the first step, a liposomal dispersion was prepared by mixing 2.12 mL of 25 mg/mL DPPC and 5 µL of 200 mg/mL cholesterol in chloroform. Then, the solution was dried using an evaporator until the dry film was obtained. The thin film of lipid was dispersed by adding 8 mL of 57 mg/mL bovine serum albumin in water, 1 mL of 15 mg/mL transferrin in water, and 88.5 µL of 10 mM ascorbate solution. The mixture was sonicated until it fully dispersed. The dispersion was adjusted to 10 mL by adding 1M HBSS). The dispersion was frozen at −20 • C for 6 h, and primary dried in the lyophilizer at −40 • C for 30 h and ramp to 25 • C for 12 h and finally secondary dried at 25 • C for 20 h. The lyophilized SLF was stored at 2-8 • C and reconstituted with purified water upon solubility and dissolution testing.

Solubility Testing
The solubility of crystalline remdesivir, amorphous remdesivir, F10 and F13 in SLF was evaluated. The excess amount of powder was added in SLF until precipitation occurred. Then the samples were shaken on the orbital shaker at 250 rpm and 25 • C for 24 h. The samples were collected and centrifuged at 13,000 rpm for 15 min. The supernatants were collected and then diluted with DMSO/acetonitrile (10/90). The samples were centrifuged again at 13,000 rpm for 15 min to separate proteins in SLF. The supernatants were collected and analyzed by HPLC, as described in Section 2.3.

Dissolution Testing
Dissolution testing of F10, F11, F12, F13, F14 and unprocessed crystalline remdesivir was performed under sink conditions in SLF and using a transwell system based upon previously published studies [33][34][35]. An NGI (MSP Corp, Shoreview, MN, USA) connected to a High-Capacity Pump (model HCP5, Copley Scientific, Nottingham, UK) and a Critical Flow Controller (model TPK 2000, Copley Scientific, Nottingham, UK) was used to load aerosolized powder. Three 24-mm diameter filter papers were placed on stage 4 on the NGI to capture aerosolized powder, which corresponded to an aerodynamic size cut off of 1.66 µm at 60 L/min. The transwell systems were modified as described in Rohrschneider et al. [35]. The filter papers were loaded in the donor compartment of the modified transwell systems. Reconstituted SLF was heated up to 37 • C before loading 1.5 mL in the six-well plate. The experiment was initiated after placing the donor compartment in the plate and adding 0.1 mL of SLF onto the filter papers. The modified transwell systems was placed on an orbital shaker at 100 rpm and 37 • C. Samples of 0.1 mL were collected at 10, 20, 30, 45, 60, 90, 120, 180, 240, 360, and 540 min. Samples were diluted with DMSO/acetonitrile (10/90) and centrifuged at 13,000 rpm for 15 min to separate proteins in SLF. Supernatants were collected and analyzed by HPLC, as described in Section 2.3. Removed volumes were refilled with fresh SLF to maintain the dissolution volume. At the last time point, the filter membranes were removed and washed with DMSO/acetonitrile (10/90) to analyze the amount of undissolved drug.

Stability Study
The TFF powder of F10, F12 and F13 was transferred to a borosilicate glass vial. The vials were dried in a lyophilizer at 25 • C and 100 mTorr for 6 h, then backfilled with nitrogen gas before stoppering inside the lyophilizer. The vials were sealed with aluminum caps before packing in an aluminum foil bag with a desiccant. All samples were stored at 25 • C/60% relative humidity (RH). After one month, samples were collected and analyzed to determine their physical stability and chemical stability. Additionally, samples were analyzed to determine aerosol performance. An in vivo pharmacokinetic study was conducted using Sprague-Dawley rats in compliance with the Institutional Animal Care and Use Committee (IACUC; Protocol Number AUP-2019-00253) guidelines at The University of Texas at Austin. Rats weighing between 200 and 250 g (average weight of 214.3 g) were housed in a 12-h light/dark cycle with access to food and water ad libitum and were subjected to one week of acclimation time to the housing environment.
Intratracheal administration was carried out using the dry powder insufflator (DP-4M model, Penn-Century Inc., Philadelphia, PA, USA) connected to the air pump (AP-1 model, Penn-Century Inc., Philadelphia, PA, USA). TFF powder was passed through a No. 325 sieve (45 µm aperture) to break down large aggregates. TFF remdesivir F10 and F13 were given to 5 rats per group. A precisely weighed quantity of sieved TFF powder was introduced into the sample chamber of the insufflator device. The dose was targeted at 10 mg/kg. Each rat was placed on its back at a 45 • angle, and its incisors were secured with a rubber band. A laryngoscope was used to visualize the trachea and insert the insufflator device into the trachea. The sieved TFF powder was actuated into the lung using 3-5 pumps (200 µL of air per pump). The mass of powder delivered was measured by weighing the chamber before and after dose actuations.
Following insufflation, whole blood was collected at each time point (5,15,30,2,4,8, and 24 h) by cardiac puncture into heparinized vials. The blood samples were centrifuged at 10,000 rpm for 3 min to obtain plasma. Lung were harvested after collecting blood at the last time point. Plasma samples and lungs were frozen and stored at −20 • C until analysis.

Analysis of Remdesivir and Its Metabolites in Plasma and Lung Tissue
Plasma samples were prepared by combining 100 µL of plasma with 100 µL of methanol that contained 100 ng/mL of the heavy labeled internal standards for remdesivir and GS-441524, followed by vortexing and centrifuging at 12,000 rpm for 15 min. The supernatant was placed in a 96-well plate for LC/MS/MS analysis.
Lung tissue samples were cut in small pieces, added into a 2 mL tube with 3.5 g of 2.3 mm zirconia/silica beads (BioSpec Producs, Bartlesville, OK, USA), and homogenized at 4800 rpm for 20 s. After homogenization, 1000 µL of methanol, containing 100 ng/mL internal standards for remdesivir and GS-441524 was added to the tube, and the tube was vortexed and centrifuged at 12,000 rpm for 15 min. The supernatant was placed in a 96-well plate for LC/MS/MS analysis. Calibration standards were prepared for plasma and lung tissue in the same fashion, spiking remdesivir and GS-441524 standard solutions into the blank plasma and lung tissue to obtain matrix matched calibrations. The calibration range was 0.1-1000 ng/mL for plasma and 50-10,000 ng/mL for lung tissue. The calibration range was chosen to bracket sample levels measured.
Remdesivir and GS-441524 were separated on an Agilent Poroshell column (2.1 × 50 mm, 2.7 µm) (Agilent, Santa Clara, CA, USA) using a gradient of 0 to 90.25% of acetonitrile with 0.025% trifluoroacetic acid in 5 min at a flow rate of 0.35 mL/min, and a column temperature of 40 • C. In total, 10 µL of each sample was injected for analysis with an Agilent 6470 triple quadrupole LC/MS/MS system (Agilent, Santa Clara, CA, USA).

Pharmacokinetic Analysis
The concentration of remdesivir and GS-441524 in plasma versus time was plotted. The data are presented as the mean ± standard deviation (SD). The plots were used to compare the pharmacokinetic behavior of F10 and F13. Microsoft Excel 2016 was used to perform a one compartmental analysis to determine key pharmacokinetic parameters including area under the plasma concentration-time curve from time 0 to 24 h (AUC 0-24h ), and area under the plasma concentration-time curve from time 0 extrapolated to infinity (AUC inf ).

Statistical Analysis
The statistical significance of experimental results was conducted using Student's t-test in JMP 10 (SAS Institute, Cary, NC, USA). The alpha level was set at 0.05.

Physical Properties of TFF Remdesivir
The particle morphology of TFF remdesivir powder formulations containing different drug loading concentrations and different excipients was determined using SEM. All formulations exhibited a brittle matrix structure of highly porous particles ( Figure 1). For acetonitrile/water solvent system, Captisol ® -and leucine-based formulations containing the same drug loading showed relatively smaller nanostructure aggregates, compared to mannitol-and lactose-based formulations. Different cosolvent systems resulted in different brittle matrix powder morphologies as observed by SEM. At the same drug load, mixture prepared from 1,4-dioxane/water produced smaller nanostructure aggregates, compared to an equivalent mixture prepared from acetonitrile/water ( Figure 1B).
Pharmaceutics 2020, 12, x 9 of 27 compartmental analysis to determine key pharmacokinetic parameters including area under the plasma concentration-time curve from time 0 to 24 h (AUC0-24h), and area under the plasma concentration-time curve from time 0 extrapolated to infinity (AUCinf).

Statistical Analysis
The statistical significance of experimental results was conducted using Student's t-test in JMP 10 (SAS Institute, Cary, NC, USA). The alpha level was set at 0.05.

Physical Properties of TFF Remdesivir
The particle morphology of TFF remdesivir powder formulations containing different drug loading concentrations and different excipients was determined using SEM. All formulations exhibited a brittle matrix structure of highly porous particles ( Figure 1). For acetonitrile/water solvent system, Captisol ® -and leucine-based formulations containing the same drug loading showed relatively smaller nanostructure aggregates, compared to mannitol-and lactose-based formulations. Different cosolvent systems resulted in different brittle matrix powder morphologies as observed by SEM. At the same drug load, mixture prepared from 1,4-dioxane/water produced smaller nanostructure aggregates, compared to an equivalent mixture prepared from acetonitrile/water ( Figure 1B).  Figure 2 shows the x-ray diffraction patterns of TFF remdesivir powder formulations. No sharp peaks of remdesivir were observed in any of formulations, indicating that remdesivir was amorphous after the TFF process. The drug loading and type of co-solvent did not affect the morphology of remdesivir. For excipients, sharp peaks of mannitol (13.5, 17, 18.5, 20.2,21, 22, 24.5, 25, 27.5, and 36 degree two-theta) were observed in F3, F7 and F11, indicating that mannitol in these formulations remained crystalline as a mixture of δ and α form [36]. Similarly, some peaks of leucine (6 and 19 degree two-theta) were observed in the TFF remdesivir-leucine formulations, F5, F9, F13, and F20, indicating that leucine remained crystalline after the process. In contrast, Captisol ® and lactose in the compositions were amorphous after the TFF process.  Figure 2 shows the x-ray diffraction patterns of TFF remdesivir powder formulations. No sharp peaks of remdesivir were observed in any of formulations, indicating that remdesivir was amorphous after the TFF process. The drug loading and type of co-solvent did not affect the morphology of remdesivir. For excipients, sharp peaks of mannitol (13.5, 17, 18.5, 20.2,21, 22, 24.5, 25, 27.5, and 36 degree two-theta) were observed in F3, F7 and F11, indicating that mannitol in these formulations remained crystalline as a mixture of δ and α form [36]. Similarly, some peaks of leucine (6 and 19 degree two-theta) were observed in the TFF remdesivir-leucine formulations, F5, F9, F13, and F20, indicating that leucine remained crystalline after the process. In contrast, Captisol ® and lactose in the compositions were amorphous after the TFF process.  Figure 2 shows the x-ray diffraction patterns of TFF remdesivir powder formulations. No sharp peaks of remdesivir were observed in any of formulations, indicating that remdesivir was amorphous after the TFF process. The drug loading and type of co-solvent did not affect the morphology of remdesivir. For excipients, sharp peaks of mannitol (13.5, 17, 18.5, 20.2,21, 22, 24.5, 25, 27.5, and 36 degree two-theta) were observed in F3, F7 and F11, indicating that mannitol in these formulations remained crystalline as a mixture of δ and α form [36]. Similarly, some peaks of leucine (6 and 19 degree two-theta) were observed in the TFF remdesivir-leucine formulations, F5, F9, F13, and F20, indicating that leucine remained crystalline after the process. In contrast, Captisol ® and lactose in the compositions were amorphous after the TFF process.  Modulated differential scanning calorimetry (mDSC) was employed to identify the glassforming ability of remdesivir and determine the glass transition temperature (Tg) of remdesivir in each formulation. Figure 3A shows a mDSC thermogram of remdesivir unprocessed powder. The first heating cycle demonstrated that remdesivir was crystalline with a melting point of ~133 °C. The cooling and second heating cycle showed that the glass transition temperature of remdesivir was about 60 °C. No recrystallization peak was observed in any cycle. Figure 3B shows the physical state of remdesivir and excipients. Despite the different formulation compositions, no melting peak of remdesivir was observed in any formulations, and it remained amorphous. These results agree with XRD diffractograms showing that remdesivir was amorphous after the TFF process. The mDSC thermograms showed that the Tg of remdesivir in all formulations was in the range 56-61 °C. Captisol ® in both F6 and F9 showed no endothermic peaks, demonstrating that Captisol ® was in amorphous form. Endothermic peak of mannitol in F7 and F11 was observed at ~165 °C and ~160 °C, respectively, indicating mannitol was crystalline after TFF process. For lactose-based formulations, two glass transition temperatures were observed, including Tg of remdesivir (~59 °C) and Tg of lactose (~117 °C), indicating lactose and remdesivir were not molecularly miscible. For leucine-based formulations, an endothermic peak was observed in F9 and F13 at ~227 °C and ~203 °C, respectively, which are not identical with the melting peak of unprocessed leucine (~300 °C) [37]. Therefore, the melting point depression of leucine can be attributed to an interaction of remdesivir and leucine. Modulated differential scanning calorimetry (mDSC) was employed to identify the glass-forming ability of remdesivir and determine the glass transition temperature (Tg) of remdesivir in each formulation. Figure 3A shows a mDSC thermogram of remdesivir unprocessed powder. The first heating cycle demonstrated that remdesivir was crystalline with a melting point of~133 • C. The cooling and second heating cycle showed that the glass transition temperature of remdesivir was about 60 • C. No recrystallization peak was observed in any cycle. Figure 3B shows the physical state of remdesivir and excipients. Despite the different formulation compositions, no melting peak of remdesivir was observed in any formulations, and it remained amorphous. These results agree with XRD diffractograms showing that remdesivir was amorphous after the TFF process. The mDSC thermograms showed that the Tg of remdesivir in all formulations was in the range 56-61 • C. Captisol ® in both F6 and F9 showed no endothermic peaks, demonstrating that Captisol ® was in amorphous form. Endothermic peak of mannitol in F7 and F11 was observed at~165 • C and~160 • C, respectively, indicating mannitol was crystalline after TFF process. For lactose-based formulations, two glass transition temperatures were observed, including Tg of remdesivir (~59 • C) and Tg of lactose (~117 • C), indicating lactose and remdesivir were not molecularly miscible. For leucine-based formulations, an endothermic peak was observed in F9 and F13 at~227 • C and~203 • C, respectively, which are not identical with the melting peak of unprocessed leucine (~300 • C) [37]. Therefore, the melting point depression of leucine can be attributed to an interaction of remdesivir and leucine. Pharmaceutics 2020, 12,
An effect of the excipient on aerosol performance was also observed. At the same drug loading, leucine-based formulations exhibited significantly greater FPF (81-85% FPF <5µm ), and a smaller MMAD (0.7-1.5 µm), compared to other excipients, indicating that leucine has distinct advantages over other compositions in terms of aerosol performance.
We also investigated the effect of co-solvent mixture and solid content on the aerosol performance of the formulations. For 100% drug loading (i.e., no excipient present in composition), significant differences in MMAD, FPF <5µm , and EF were observed between the formulations prepared from the acetonitrile/water (F14) and 1,4-dioxane/water (F15 and F16) co-solvent systems. F15 and F16 exhibited a smaller MMAD, and a larger FPF and EF compared to F14. Comparing the formulations prepared in the same solvent system, F16 showed a better aerosol performance, although it was prepared at higher solid content.
An effect of the excipient on aerosol performance was also observed. At the same drug loading, leucine-based formulations exhibited significantly greater FPF (81-85% FPF< 5 µm), and a smaller MMAD (0.7-1.5 µm), compared to other excipients, indicating that leucine has distinct advantages over other compositions in terms of aerosol performance.
We also investigated the effect of co-solvent mixture and solid content on the aerosol performance of the formulations. For 100% drug loading (i.e., no excipient present in composition), significant differences in MMAD, FPF< 5 µm, and EF were observed between the formulations prepared from the acetonitrile/water (F14) and 1,4-dioxane/water (F15 and F16) co-solvent systems. F15 and F16 exhibited a smaller MMAD, and a larger FPF and EF compared to F14. Comparing the formulations prepared in the same solvent system, F16 showed a better aerosol performance, although it was prepared at higher solid content.

Interactions Between TFF Remdesivir and Excipients
1 H-NMR was performed to identify interactions between remdesivir and excipients. Figure 5B demonstrates an expansion of 1 H-NMR spectra for selected TFF remdesivir powder formulations and remdesivir unprocessed powder. While the peak at 6.03 ppm is sharp and does not show any differences from the presented samples, the peaks at 5.37 and 6.33 ppm of F9 were broader. Moreover, these peaks were slightly shifted downfield. Figure 5C shows an expansion of the 1 H-NMR spectra of F9 and leucine unprocessed powder. The peak at 3.09 ppm of leucine was also shifted downfield to 3.13 ppm in F9. These results indicate that remdesivir and leucine form interactions during the TFF process.

Chemical Stability of TFF Remdesivir Compositions
The chemical stability of remdesivir after TFF processing was confirmed by 1 H-NMR. Figure 5A presents the 1 H-NMR spectra of selected TFF remdesivir powder formulations and remdesivir unprocessed powder. Remdesivir peaks from the TFF remdesivir powder formulations consisting of 50% (w/w) remdesivir with mannitol, lactose, and leucine (F7, F8, and F9, respectively) are identical to the peaks of remdesivir unprocessed powder. The 80% (w/w) remdesivir with Captisol ® (F10) and 100% remdesivir without excipient (F14) powders also presented indistinguishable peaks corresponding to remdesivir. Therefore, neither the aqueous-organic cosolvent mixture nor the conditions used in the TFF process affected the chemical stability of remdesivir.
After one month of storage at 25 °C/60% RH, samples were analyzed for chemical stability by NMR and HPLC. NMR spectra demonstrated that no chemical shift of remdesivir peaks and no new peaks were observed in F10, F12 and F13. Additionally, no degradant peak was observed in the HPLC chromatograms (data not shown), which agrees with the NMR spectra. Both analyses indicated that remdesivir was chemically stable without degradation after storage.

Physical Stability of TFF Remdesivir Compositions
Samples after storage at 25°C/60% RH were analyzed for physical stability by XRD. The XRD diffractogram demonstrates that no remdesivir peaks were observed in any samples after storage (Figure 6), indicating that remdesivir in F10, F12 and F13 remained amorphous after one month of

Chemical Stability of TFF Remdesivir Compositions
The chemical stability of remdesivir after TFF processing was confirmed by 1 H-NMR. Figure 5A presents the 1 H-NMR spectra of selected TFF remdesivir powder formulations and remdesivir unprocessed powder. Remdesivir peaks from the TFF remdesivir powder formulations consisting of 50% (w/w) remdesivir with mannitol, lactose, and leucine (F7, F8, and F9, respectively) are identical to the peaks of remdesivir unprocessed powder. The 80% (w/w) remdesivir with Captisol ® (F10) and 100% remdesivir without excipient (F14) powders also presented indistinguishable peaks corresponding to remdesivir. Therefore, neither the aqueous-organic cosolvent mixture nor the conditions used in the TFF process affected the chemical stability of remdesivir.
After one month of storage at 25 • C/60% RH, samples were analyzed for chemical stability by NMR and HPLC. NMR spectra demonstrated that no chemical shift of remdesivir peaks and no new peaks were observed in F10, F12 and F13. Additionally, no degradant peak was observed in the HPLC chromatograms (data not shown), which agrees with the NMR spectra. Both analyses indicated that remdesivir was chemically stable without degradation after storage.

Physical Stability of TFF Remdesivir Compositions
Samples after storage at 25 • C/60% RH were analyzed for physical stability by XRD. The XRD diffractogram demonstrates that no remdesivir peaks were observed in any samples after storage (Figure 6), indicating that remdesivir in F10, F12 and F13 remained amorphous after one month of storage.

Physical Stability of TFF Remdesivir Compositions
Samples after storage at 25°C/60% RH were analyzed for physical stability by XRD. The XRD diffractogram demonstrates that no remdesivir peaks were observed in any samples after storage (Figure 6), indicating that remdesivir in F10, F12 and F13 remained amorphous after one month of storage.

Aerosol Performance after Storage
The aerosol performance of F10, F12 and F13 was analyzed after one month of storage at 25 • C/60% RH (Table 3 and Figure 7). Overall, MMAD and FPF <5µm (of recovered dose) of F10, F12, and F13 was in the range of 1-2 µm and 74-83%, respectively (Table 3). Although MMAD was slightly decreased in all three formulations, there was no significant change in the aerosol performance after one month of storage at 25 • C/60% RH. Table 3. Aerosol performance of TFF remdesivir powders after one month of storage at 25 • C/60% RH using a Plastiape ® RS00 high-resistance DPI at a flow rate of 60 L/min (n = 3). (MMAD, mass median aerodynamic diameter; GSD, geometric standard deviation; FPF, fine particle fraction; EF, emitted fraction).

Solubility in Simulated Lung Fluid
The solubility of neat remdesivir in SLF was evaluated. The solubility of amorphous remdesivir was about 20 times higher than that of crystalline remdesivir (0.59 ± 0.01 mg/mL vs. 0.03 ± 0.001 mg/mL). The solubility of remdesivir in F10 and F13 was also determined. The solubility of F10 and F13 was 0.60 ± 0.04 mg/mL and 0.57 ± 0.03 mg/mL, respectively, which was similar to the solubility of amorphous neat remdesivir.

Drug Release in Simulated Lung Fluid
The in vitro dissolution profiles of remdesivir in SLF media under sink conditions are shown in Figure 8. The dissolution rates of TFF remdesivir formulations were significantly higher than the unprocessed crystalline remdesivir (p < 0.05). More than 80% of remdesivir in F10, F11, F12 and F13 was released and dissolved within 90 min, while 80% remdesivir in F14 was dissolved in 180 min.

Solubility in Simulated Lung Fluid
The solubility of neat remdesivir in SLF was evaluated. The solubility of amorphous remdesivir was about 20 times higher than that of crystalline remdesivir (0.59 ± 0.01 mg/mL vs. 0.03 ± 0.001 mg/mL). The solubility of remdesivir in F10 and F13 was also determined. The solubility of F10 and F13 was 0.60 ± 0.04 mg/mL and 0.57 ± 0.03 mg/mL, respectively, which was similar to the solubility of amorphous neat remdesivir.

Drug Release in Simulated Lung Fluid
The in vitro dissolution profiles of remdesivir in SLF media under sink conditions are shown in Figure 8. The dissolution rates of TFF remdesivir formulations were significantly higher than the unprocessed crystalline remdesivir (p < 0.05). More than 80% of remdesivir in F10, F11, F12 and F13 was released and dissolved within 90 min, while 80% remdesivir in F14 was dissolved in 180 min.

In Vivo Pharmacokinetic Study
The plasma concentration-time profiles of remdesivir and GS-441524 following a single inhalation dose of TFF remdesivir formulations are shown in Figure 9A,B, respectively. Remdesivir was detected in plasma at 5 min (0.068 ng/mL) and reached a maximum 30 min (1.149 ng/mL) following pulmonary dosing of F10. A much smaller amount of remdesivir (< 0.01 ng/mL) was detected at 5 and 15 min following pulmonary dosing of F13. Remdesivir concentrations were below the lower limit of quantitation in plasma after 2 h and after 30 min following pulmonary dosing of F10 and F13, respectively. Figure 9B shows a comparison of plasma GS-441524 concentration-time profile from F10 and

In Vivo Pharmacokinetic Study
The plasma concentration-time profiles of remdesivir and GS-441524 following a single inhalation dose of TFF remdesivir formulations are shown in Figure 9A,B, respectively. Remdesivir was detected in plasma at 5 min (0.068 ng/mL) and reached a maximum 30 min (1.149 ng/mL) following pulmonary dosing of F10. A much smaller amount of remdesivir (<0.01 ng/mL) was detected at 5 and 15 min following pulmonary dosing of F13. Remdesivir concentrations were below the lower limit of quantitation in plasma after 2 h and after 30 min following pulmonary dosing of F10 and F13, respectively. detected at 5 and 15 min following pulmonary dosing of F13. Remdesivir concentrations were below the lower limit of quantitation in plasma after 2 h and after 30 min following pulmonary dosing of F10 and F13, respectively. Figure 9B shows a comparison of plasma GS-441524 concentration-time profile from F10 and F13, and the calculated pharmacokinetic parameters following pulmonary administration are presented in Table 4. The plasma GS-441524 profiles following pulmonary dosing of F10 and F13 has peak concentrations (Cmax) of 220.4 ± 118.0 ng/mL at 2 h and 264.3 ± 88.5 ng/mL at 4 h, respectively, before plasma and GS-441524 that remained in the lung tissues of treated animals at 24 h following pulmonary administration of F10 and F13. Despite no statistical difference, F10 exhibited greater amounts of remdesivir in the lungs at 24 h, while F13 showed greater amounts of GS-441524 in the lungs at 24 h.    Figure 9B shows a comparison of plasma GS-441524 concentration-time profile from F10 and F13, and the calculated pharmacokinetic parameters following pulmonary administration are presented in Table 4. The plasma GS-441524 profiles following pulmonary dosing of F10 and F13 has peak concentrations (C max ) of 220.4 ± 118.0 ng/mL at 2 h and 264.3 ± 88.5 ng/mL at 4 h, respectively, before Pharmaceutics 2020, 12, 1002 20 of 28 plasma concentrations started decreasing. The AUC 0-24 and AUC inf of F13 were significantly greater than those of F10 (p < 0.05).  Table 5 shows the amount of remdesivir and GS-441524 that remained in the lung tissues of treated animals at 24 h following pulmonary administration of F10 and F13. Despite no statistical difference, F10 exhibited greater amounts of remdesivir in the lungs at 24 h, while F13 showed greater amounts of GS-441524 in the lungs at 24 h.

Thin Film Freezing Produces High Potency Remdesivir Dry Powders for Inhalation with High Aerosol Performance
We investigated the feasibility of using the thin film freezing process to prepare inhalable remdesivir powder formulations. Different excipients, including Captisol ® , mannitol, lactose, and leucine, were evaluated in this study. Mannitol and lactose are presently contained in FDA approved inhalation dosage forms, while leucine has gained recent interest for pulmonary delivery and is being used in an inhaled product in FDA clinical trials [38][39][40]. Captisol ® was selected in this study since it is a solubilizer used in both the solution concentrate for dilution and infusion and the lyophilized powder for reconstitution/dilution and infusion of current remdesivir products [3].
The RS00 high-resistance monodose DPI is a capsule-based DPI device that is available for commercial product development, and it functions to disperse the powder based on impaction force. A previous study confirmed that this impact-based DPI can disperse low-density brittle matrix powders made by TFF process into respirable particles better than a shear-based DPI (e.g., Handihaler ® ) [41]. Another study also evaluated the performance of different models of the monodose DPI (RS01 and RS00) on the aerosol performance of brittle matrix powders containing voriconazole nanoaggregates prepared by TFF [42]. It was shown that the RS00 device exhibited better powder shearing and deaggregation through smaller holes of the capsule created by the piercing system of the RS00 device [42]. Therefore, the RS00 high-resistance Plastiape ® DPI was selected for this study.
We found that excipient type and drug loading affect the aerosol performance of TFF remdesivir powder compositions. Overall, the aerosol performance of TFF remdesivir powders increased as the drug loading was increased, a highly desirable feature. This trend is obvious for the Captisol ® -, lactose-, and mannitol-based formulations when the drug loading was increased from 20% to 50%. Furthermore, high-potency TFF remdesivir powder without excipients (F14 and F15) also exhibited high FPF and small MMAD, which indicates that remdesivir itself has a good dispersing ability without the need of a dispersing excipient when prepared using the TFF process. This shows that the TFF technology can be used to minimize the need of excipient(s) in the formulation, thus maximizing the amount of remdesivir being delivered to the lungs by dry powder inhalation.
The aerosol performance of the leucine-based formulations did not significantly change when the drug loading was increased from 20% to 80%, and these formulations exhibited superior aerosol performance compared to the other excipient-based formulations studied in this paper. This is likely attributed to the surface modifying properties of leucine. Several papers report that leucine can minimize the contact area and distance between particles [43,44]. This decreases Van der Waal forces between drug particles and subsequently increases aerosol performance.
Additionally, different co-solvent systems affected the aerosol performance of TFF remdesivir powders. The formulations, except for the TFF remdesivir-leucine, prepared in a 1,4-dioxane/water co-solvent system exhibited smaller MMAD and higher FPF than the formulations prepared in an acetonitrile/water co-solvent system. Interestingly, the co-solvent systems did not affect the aerosol performance of the TFF remdesivir-leucine formulation. This agrees with SEM figures showing that the formulations prepared in a 1,4-dioxane/water co-solvent system have smaller nanostructured aggregates than the formulations prepared in an acetonitrile/water cosolvent system. This may be due to the difference in the viscosity of the solvent system. The viscosity of acetonitrile/water (50/50 v/v) was lower than that of 1,4-dioxane/water (50/50 v/v) (0.81 vs. 1.62 mPa·s) [45,46]. Our results agree with previous studies by Beinborn et al. [47] and Moon et al. [42] showing that the viscosity of the solvent system has an impact on the aerosol performance of TFF powder. The higher viscosity of the co-solvent minimizes the movement of molecules during the ultra-rapid freezing step, resulting in more homogenous distribution in the frozen state [42]. On the contrary, the lower-viscosity solvent allows molecules to move more readily, which increases the chance of molecular aggregation and subsequently decreases the aerosol performance [42].

Physical and Chemical Stability of Remdesivir Dry Powder Produced Using TFF
Both XRD diffractograms and DSC thermograms showed that remdesivir was amorphous after the TFF process. An amorphous form of the drug generally provides for faster dissolution rate than its crystalline form. Since the drug needs to be absorbed and then penetrate through the cell membrane before it is hydrolyzed to nucleoside monophosphate GS-441524, TFF remdesivir powders provide benefits for the dissolution of the deposited powder in the lung fluid that leads to improvement in the bioavailability and efficacy of the drug when administered by inhalation.
One potential concern related to the amorphous drug is physical instability due to its high-energy state. According to criteria described by Wyttenbach et al., our study confirmed that remdesivir is categorized as a class III glass-forming drug (i.e., it is a stable glass former) [48], because no crystallization peak was observed in both the cooling and heating cycles on DSC. It was reported that the glass-forming ability (GFA) has an influence on the physical stability of drug [49]. The GFA class III drugs have a strong tendency to transform into its amorphous state, resulting in the highest physical stability compared to class I and class II active pharmaceutical ingredients [49,50].
Additionally, NMR spectra demonstrated that Captisol ® and leucine have interactions with remdesivir, which may help to stabilize remdesivir during storage. Moreover, DSC thermograms indicated that lactose and remdesivir were not molecularly dispersed in one single amorphous phase. To investigate the effects of the drug and excipient interactions on physical and chemical stability, we selected F10, F12 and F13 in the stability study.
We found that the properties of a class III glass-forming drug agree with the physical stability of remdesivir. After one month of storage at 25 • C/60% RH, the results showed that remdesivir in all three formulations was physically stable. Different types of excipients did not affect the physical stability of remdesivir. Even though lactose was not molecularly dispersed with remdesivir, its use did not affect the physical stability of remdesivir. Lastly, these results indicate that amorphous remdesivir was physically stable without a stabilizer.
In addition to the physical stability, remdesivir, as a prodrug, is prone to degrade by hydrolysis in aqueous solution. Since an organic/water co-solvent system is required to dissolve the drug and excipients in the TFF process, chemical stability is another concern during preparation. NMR spectra demonstrated that remdesivir did not chemically degrade as a result of the TFF process. Even though remdesivir was exposed to binary co-solvent systems consisting of water during the process, the entire TFF process used to produce remdesivir dry powder inhalation formulations did not induce chemical degradation of the parent prodrug. Furthermore, remdesivir was chemically stable by HPLC (data not shown) and NMR after one month of storage at 25 • C/60% RH.

Remdesivir in TFF Powder Compositions Can Be Dissolved, Absorbed, and Metabolized to GS-441524 in the Lungs
Since remdesivir is a poorly water-soluble drug, its dissolution may be a critical factor of drug release in the lung fluid, especially in high drug load formulations. Undissolved particles can be cleared by mucociliary clearance or macrophage uptake, causing lower drug concentration, lung irritation, and inflammatory response [51]. Therefore, a dissolution test was evaluated in this study. The dissolution profile demonstrated that physical form of drug appears to have a significant effect on the dissolution rate. Amorphous remdesivir prepared using TFF had a much faster dissolution rate in simulated lung fluid compared to crystalline remdesivir. This agrees with the solubility of remdesivir. Amorphous neat remdesivir showed a 20-fold increase in solubility compared to crystalline neat remdesivir. Moreover, the high porosity and high surface area of the TFF powders also contribute to faster wetting, thereby enhancing the dissolution rate [31,52].
The presence of excipients also affected the dissolution rate of remdesivir. The dissolution rate of 100% drug load formulation (i.e., no excipient was used), even though it is amorphous, was slower, compared to 80% drug loading formulations. The enhancement of dissolution rate is likely attributed to improved wetting from the use of excipients. Lactose and mannitol are hydrophilic carriers that can wet the powder formulations rapidly, promoting faster exposure of the drug particles to the dissolution medium [53]. In the case of Captisol ® , cyclodextrins and their derivatives have been widely used as solubility and dissolution modifying excipients. In our study, the molar ratio of remdesivir to Captisol ® in the high drug load composition (80/20 w/w) was 10:1, meaning that only about 10% of remdesivir was present in a complex with Captisol ® in the solubilized form; the remaining remdesivir was not complexed with Captisol ® and was present as amorphous powder in the matrix. It is known that certain drugs form a complex with Captisol ® by hydrophobic interactions [15][16][17]. The inclusion complexation between cyclodextrin and a drug can generally increase the aqueous solubility of the aforementioned drug, and increase the dissolution rate [54]. It was previously reported that the solubility of remdesivir can be increased by Captisol ® , and Veklury ® , and the commercial product containing remdesivir includes Captisol ® to solubilize remdesivir for the IV solution formulation [15]. However, in our study, 10% complexation of remdesivir and Captisol ® in F10 appeared to have an effect on dissolution rate, but it did not affect the solubility of remdesivir since there was no significant difference in the solubility between F10 and amorphous neat remdesivir.
Interestingly, the TFF leucine-based formulation exhibited a higher dissolution rate than the TFF remdesivir formulations that contain mannitol or lactose, despite its hydrophobic properties. The faster dissolution rate of the leucine-based formulation is likely attributed to interactions (e.g., hydrogen bonding) between remdesivir and leucine. Peaks corresponding to acidic protons (-OH or -NH) of remdesivir from F9, a leucine-based formulation, became broader than those from the other formulations without leucine, and shifted downfield. These wider peaks result from inter-molecular interactions [55]. A peak shift in NMR spectra indicates magnetic field changes. The downfield shift of the acidic protons from F9 is caused by deshielded protons and reduced electron density, indicating hydrogen bonding at the protons [56,57]. Along with a leucine-corresponding peak shift from F9, it was concluded that remdesivir favorably interacts with leucine in hydrogen bonding. The molar ratio of remdesivir to leucine of the high drug load TFF compositions (80/20 w/w) was 1:1.5, meaning that leucine in this composition was in excess in terms of its ability to interact with remdesivir. Similarly, leucine interactions have been reported in Mangal's study to enhance the dissolution rate of azithromycin [58]. Specifically, leucine formed a complex with azithromycin through hydrogen bonding [58]. Generally, even though leucine is a hydrophobic amino acid, it is a water-soluble excipient that has higher solubility than poorly water-soluble drugs like remdesivir. Therefore, the interactions between drug and leucine at the molecular level can increase the dissolution rate of poorly water-soluble drugs [58,59], like the one observed in our study.
Significantly higher plasma levels of GS-441524 than remdesivir were observed in our rat pharmacokinetic study. The half-life of remdesivir is reportedly much shorter than that of GS-441524. While the half-life of GS-441524 is approximately 24.5 h, for remdesivir it is only about 1 h in humans following multiple IV administrations once a day [3]. In rats, the half-life of remdesivir in plasma is reported to be less than 0.9 min, which is much shorter than that in humans, due to higher esterase activity in rodent plasma [3]. Therefore, with a T max of 30 min and a level of remdesivir in plasma that is lower than the detection limit at 2 h, F10 presented an initial rapid systemic absorption of remdesivir that was complete at less than 2 h. T max of GS-441524 plasma concentration (2 h) also supports this observation. In comparison, remdesivir plasma concentrations of F13 were lower than those of F10 at 5 and 15 min, and it was lower than the detection limit at the other time points, indicating that the initial systemic absorption of remdesivir from lung was much slower with F13. With the later T max of GS-441524 plasma concentration, however, more sustained release of remdesivir from lung to plasma was observed.
The initial faster systemic absorption of F10 from the lungs is likely due to the fact that F10 contains about 10% remdesivir (on a molar basis) that is complexed with Captisol ® in a solubilized form. Higher concentrations of both remdesivir and GS-441524 were detected in plasma following administration of F10, which contains 20% w/w of Captisol ® . Captisol ® (i.e., sulfobutylether-β-cyclodextrin), is known to enhance the stability, solubility, and bioavailability of drugs [60]. When Captisol ® forms a complex by hydrophobic interactions with drugs, these interactions are even greater than with simple physical mixtures of drug and Captisol ® [16,17]. Although F10 and F13 consist of amorphous remdesivir in the TFF remdesivir powder formulations, the inclusion of Captisol ® in F10 likely increased the absorption of the complexed remdesivir into systemic circulation due to its improvement in solubilization [15,61]. Even though F10 incorporates a high molar ratio of drug to Captisol ® (10:1), and not all of the remdesivir in F10 complexes with Captisol ® , a partial remdesivir-Captisol ® complex will still dissolve faster in lung fluid, rapidly becoming a solution and being absorbed into systemic circulation more quickly than the non-complexed remdesivir fraction with the Captisol ® of F10. Eriksson et al. described that drugs in solution can be absorbed faster than dry powder in lungs in the isolated perfused lung model [62]. Enhanced absorption of inhaled drugs complexed with Captisol ® from the lungs to systemic circulation was reported in inhaled voriconazole studies. A solution form of nebulized voriconazole formulation with Captisol ® resulted in an eightfold higher AUC plasma to AUC lung ratio compared to nanostructured amorphous voriconazole powder formulations when delivered to the lung directly (0.64 vs. 0.08, respectively) [22,24].
Lastly, and related to potential use in COVID-19 therapy, the 50% maximal effective concentration (EC 50 ) value is important to evaluate the antiviral efficacy of drugs. While there are studies reported to evaluate remdesivir potency to inhibit SARS-CoV-2 replication, only one study has reported antiviral efficacy of GS-441524 against SARS-CoV-2. Pruijssers et al. determined the inhibition of SARS-CoV-2 replication in established cell lines by GS-441524, and reported that EC 50 is 0.62 µM (180.6 ng/mL) in Calu3 2B4 cells, and 0.47 µM (136.9 ng/mL) in Vero E6 cells [63]. The target dose of 10 mg/kg by dry powder insufflation in our rat pharmacokinetic study achieved the EC 50 values necessary to reach EC 50 in order to inhibit SARS-CoV-2. The plasma concentrations of GS-441524 in our pharmacokinetic study also achieved higher than the reported EC 50 of 0.18 µM (52.4 ng/mL) in SARS-CoV-infected human airway epithelial cells [64].

Conclusions
We have demonstrated that low-density, highly porous brittle matrix particles of remdesivir powder formulations made by thin film freezing are highly aerosolizable and suitable for inhaled delivery to the lung, the site of SARS-CoV-2 replication. The properties of a good glass former, like remdesivir, supports that remdesivir maintains its physical stability during storage if protected from moisture. The thin film freezing process is suitable to prepare remdesivir dry powder for inhalation without causing chemical degradation. Dry powder administration can deliver remdesivir to the lungs, which appears to be converted to its nucleoside analog GS-441524 in the lungs before entering system circulation based on the data showing negligible levels of remdesivir in the blood at all times for the TFF remdesivir-leucine formulation. Therefore, dry powder remdesivir for inhalation produced using thin film freezing technology can provide a dry powder inhalation product for the treatment of patients with COVID-19 that can be administered on an outpatient basis and earlier in the disease course where effective antiviral therapy has a greater potential to reduce the morbidity and mortality associated with the disease.