1. Introduction
Pulmonary delivery by inhalation is the preferred method of administration for the treatment of lung diseases such as asthma, COPD, pulmonary arterial hypertension, and pulmonary infections [
1,
2]. Inhaled treatments for additional indications such as lung cancer [
3] are in clinical trials. Combination therapies delivered by dry powder inhaler (DPI) are of particular interest for improving patient experience and compliance while managing complex lung conditions [
4]. To date, at least seven DPI combination therapies have been approved by the FDA (Breo Ellipta, Anoro Ellipta, Trelegy Ellipta, Utibro Breezhaler, Advair Diskus, Airduo Respiclick/Digihaler, Wixela Inhub (an Advair generic)) [
5]. In all of these products, the active particles are reduced to a respirable particle size by a milling step and are then blended with carrier particles. This particle engineering approach is limited to small molecule therapies which are solid, crystalline, and insensitive to shear forces sustained during a milling process [
6].
In contrast, spray drying is an enabling particle engineering technique, used to manufacture respirable particles of both small and large molecule active pharmaceutical ingredients (API) in a single-unit operation [
7,
8,
9]. Spray drying as a pharmaceutical manufacturing technique has been reviewed extensively elsewhere [
10,
11]. Briefly, actives and excipients are co-dissolved in a volatile solvent and then atomized into droplets that are sprayed into a drying chamber. Heated drying gas rapidly removes the solvent, resulting in a dried powder that is collected via cyclone. The spray-drying manufacture of an inhalation formulation with more than one active compound remains challenging. Two or more APIs can be combined into a single spray solution and spray-dried such that they are molecularly mixed. Spraying together can be challenging due to the need for a common solvent, particularly when combining a biotherapeutic (which must be dissolved in an aqueous buffer to maintain biologic activity) with a low-solubility small molecule API (e.g., BCS class II or IV) that can have very low solubility in aqueous systems. In certain cases, APIs with a common solvent will not be chemically stable in the same solution or particle. Alternatively, each active could be formulated and spray-dried separately and then blended together. This approach introduces additional processing steps, potential difficulties in content uniformity, and is particularly challenging for inhalation powders, which have very poor flow and are often hygroscopic.
Another option would be to simultaneously spray two different formulation solutions into the same spray dryer through separate atomizers—a concept demonstrated by Snyder et al., who used two phosphate buffer solutions as model systems [
12]. Numerous studies have demonstrated complex particle morphologies which combine two or more actives by spray-drying emulsions [
13], nanoparticle suspensions [
14], or other atomization technologies [
15]. In contrast, simultaneous spray drying produces two distinct particle types that are intimately blended during the spray-drying process.
This study aims to demonstrate this process, hereafter termed simul-spray drying, for the first time as a method of manufacturing combination inhalation products with matching particle size distributions for DPI. More specifically, we produced spray-dried inhalation powders consisting of both a biologic and a small molecule API that also require different spray solvents. A custom atomizer wand for a lab-scale spray dryer was fabricated to accommodate two independent two-fluid atomizers, from which two solutions could be sprayed. A schematic is provided in
Figure 1. The relative amounts of each spray-dried formulation present in the final product were controlled by the liquid flow rate to each atomizer and the composition of the solutions.
The model systems chosen for this study were selected due to potential relevance to the treatment of lung cancer: bevacizumab (BEV), a VEGF-inhibitor monoclonal antibody (mAb) first marketed as Avastin [
16], and three small molecules commonly used in tandem with BEV treatments—erlotinib (ERL, an EGFR inhibitor), paclitaxel (PTX, a chemotherapy), and cisplatin (CP, a chemotherapy). Simul-spray drying successfully generated combination spray-dried powders with appropriate aerosol properties for lung delivery and no impact on the anti-VEGF activity of BEV. The inhalable combination formulations of these actives would not be possible by other manufacturing techniques due to bevacizumab’s incompatibility with a high-shear milling process and the lack of a common spray-drying solvent between the actives.
2. Materials and Methods
2.1. Materials
The bevacizumab drug substance was supplied as a sterile solution containing 30 mg/mL bevacizumab, 60 mg/mL trehalose, and 0.04% polysorbate 20 in 50 mM phosphate buffer at pH 6.2. Trehalose dihydrate was purchased from Pfanstiehl (Waukegan, IL, USA), and L-leucine was purchased from J.T. Baker Inc. (Phillipsburg, NJ, USA). Cisplatin was purchased from BOC Sciences (Shirley, NY, USA), paclitaxel and erlotinib were purchased from LC Laboratories (Woburn, MA, USA).
2.2. Spray Drying
Solutions for the spray drying of paclitaxel, cisplatin, or erlotinib were prepared by adding API and excipient solids to the solvent and stirring until dissolved. Due to their chemical instability in aqueous solutions, cisplatin solutions were used as soon as possible after preparation. The compositions are summarized in
Table 1.
For the bevacizumab solution preparation, a dialysis buffer exchange was performed, as described a previous publication [
17].
A customized atomizer that was fabricated for this study could introduce two solutions into the dryer simultaneously through independent nozzles. A schematic is shown in
Figure 1. The atomizers are angled slightly out from one another (5–10°) such that plume interference is reduced, helping to prevent the collision and fusion of atomized droplets. The solution compositions and liquid flow rates used to manufacture the formulations in this study are listed in
Table 1 and
Table 2. The liquid streams were pumped into a pre-heated spray dryer with a nominal nitrogen drying gas flow rate of 500 g/min. A two-fluid nozzle was used for the atomization of each stream (Model ¼ J, with a 1650 liquid body and 64 air cap, Spraying Systems Co., Wheaton, IL, USA). The outlet temperature was set at 50 °C, and the inlet temperature was approximately 110 °C. Atomization gas pressures ranged from 15–20 psi. Spray-dried particles were collected using a 2 inch cyclone separator.
2.3. Drug Concentration Measurement
The drug concentration of the BEV/ERL and BEV/PTX spray-dried powders was measured using HPLC. A known mass of sample was dissolved in DMSO with sonication and analyzed on an Agilent 1100 (Agilent Technologies, Santa Clara, CA, USA), with detection at 280 nm, and then quantified against the linear standard curves of each respective active (0.05–1 mg/mL). A test solution containing known amounts of all three actives was used to confirm the specificity and accuracy of the method. A gradient method was used with a mobile phase flow rate of 1.5 mL/min and a 5 µL sample injection using an Agilent Poroshell 300 SB-C3 column at 75 °C (2.1 × 75 mm with 5 µm particles). Mobile phase A was 0.1% TFA in water, and mobile phase B was 0.1% TFA in acetonitrile, starting with 98:2 A:B for 0.1 min, then a gradient to 40:60 from 0.1 to 2 min, followed by a 0.6 min isocratic hold at 40:60 A:B before a re-equilibration back to 98:2. The total method run time including the re-equilibration step was 4.5 min, with elution of ERL, PTX, and BEV at 1.4, 1.8, and 2.1 min, respectively. ERL samples were measured in triplicate, and PTX samples were measured in quintuplicate.
BEV/CP spray-dried powders were quantitated using the 2nd derivative of the absorbance spectrum. A known mass of sample was dissolved in DMSO with sonication and analyzed using fiber optic UV-vis probes with 5 mm tips from Pion Inc. (Billerica, MA, USA) with the Au PRO software. BEV was quantitated against a linear standard curve (14–220 μg/mL) at 282 and 298 nm, identified using the software’s zero intercept mode (ZIM) as the wavelengths where CP shows no signal in the second derivative, irrespective of concentration. CP was quantitated against a linear standard curve (14–220 μg/mL) at 335–345 nm, where BEV has no absorbance. The second derivative spectra are shown in the
Supplementary Information. A test solution containing known amounts of both actives was used to confirm the specificity and accuracy of the method. All samples were measured in triplicate.
2.4. Water Content
The water content of spray-dried formulations was quantified by Karl Fischer titration on a coulometric Metrohm 851 Titrando KF oven titrator (Metrohm USA Inc., Tampa, FL, USA). The generator electrode was operated in diaphragm-less mode. Sample sizes of 10–40 mg were sealed in a crimped KF vial and analyzed at an oven temperature of 105 °C and then measured in duplicate.
2.5. Scanning Electron Microscopy (SEM)
SEM images of spray-dried formulations were obtained using a Hitachi SU3500 SEM (Hitachi High Technologies America Inc., Schaumburg, IL, USA). Trace quantities of powder were applied to a double-sided carbon tape mounted on an aluminum stud. Samples were sputter coated with gold/palladium for approximately 6 min at 15–20 mA plasma current using a Hummer 6.2 Sputter System (Anatech+ Ltd., Battle Creek, MI, USA).
2.6. Thermal Analysis by Differential Scanning Calorimetry (DSC)
Thermal analysis was performed on samples using a Mettler Toledo DSC 3+ instrument (Mettler Toledo, Columbus, OH, USA). Samples were sealed in 40 µL aluminum pans, which were vented to allow moisture to boil off during the run. Samples were scanned in ADSC mode from 0 to 160 °C at 2.5 °C/min, with a modulation of 1.5 °C in amplitude every 60 s. The glass transition temperature of the samples was quantified in the Mettler Toledo STARe software from the midpoint temperatures of the transitions, measured in triplicate.
2.7. Aerosol Properties by Fast-Screening Impactor (FSI)
The aerosol properties of the spray-dried powders were analyzed using a fast-screening impactor device (Copley Scientific, Nottingham, UK). Based on the same technology as the pre-separator of a Next Generation Impactor, the FSI separates powders into coarse and fine particle fractions. A total of 10 mg of powder was loaded into a size 3 V Caps Plus DPI grade capsule and placed in a PlastiApe RS01 4 kPa dry powder inhaler. Using a Copley HCP5 pump and a TPK2000 unit, the test was operated at 60 L/min for 2.0 s. Particles with an aerodynamic diameter of <5 microns bypassed the impaction stage and were collected on a glass filter. The mass change of the filter before and after actuation of the DPI device was used to calculate the fine particle fraction.
2.8. Aerosol Particle Size Measurement
The aerodynamic particle size distribution of spray-dried powders was quantified by a TSI Aerodynamic Particle Sizer® 3321 with a Model 3433 small-scale powder disperser and a Model 3302 A diluter (TSI, Shoreview, MN, USA). The air flow rate was 18.5 L/min in the powder disperser, and the sheath flow rate was 4 L/min. A 100:1 capillary was used in the diluter at a pressure of 0.32 in. of water. Distributions were measured in triplicate for each sample at 20 s each.
2.9. Residual Solvent Quantitation by Gas Chromatography (GC) Headspace
The concentration of residual solvent from spray dying in the powder was quantified by GC headspace. A known mass of sample was dissolved in 4 mL of dimethylacetamide in a 20 mL headspace vial. GC was performed on the headspace using an Agilent G7890 (Agilent Technologies, Santa Clara, CA, USA) equipped with a flame ionization detector and split injection capability as well as an Agilent 7697 automated headspace sampler. An Agilent DB-624 column was used with 30 m × 0.32 mm × 1.8 microns. The headspace sampler and instrument parameters used for the experiments are summarized in the
Supplementary Information.
2.10. VEGF Binding ELISA Assay
To assess the in vitro binding of spray-dried BEV simul-sprays to VEGF, a Human VEGF Quantikine ELISA kit (Part DVE00, R&D Systems, Minneapolis, MN, USA) was repurposed for a competitive ELISA assay. As a control, the as-received BEV solution was diluted to a 4 mg/mL active in 0.01 M pH 7.4 phosphate buffered saline. Spray-dried powders or as-received API were reconstituted to a 4 mg/mL active in the buffered saline and allowed to dissolve for 60 min. All samples were then centrifuged at 10,000× g for 1 min to remove undissolved solids (as would be expected for simul-sprays containing low-solubility actives such as PTX and ERL). Samples were then prepared containing 2.5 nM human recombinant VEGF (R&D Systems, Minneapolis, MN, USA) and 0.75 mg/mL BEV (from spray-dried powder or stock) in the ELISA kit’s calibrator diluent RD5K. For controls which did not contain bevacizumab, identical volumes were transferred. Samples were incubated for 60 min at 37 °C to allow BEV to bind VEGF to equilibrium. Standards were prepared according to the kit protocol.
After incubation, 50 µL of the assay diluent was added to each of the wells of the supplied plate. A total of 200 µL each of the sample (in triplicate) and standards were added to respective wells and incubated for 2 h at ambient temperature. The plate was washed three times with 400 µL of the kit’s wash buffer, then 200 µL of the VEGF-conjugate was added. The plate was incubated for another 2 h, then washed again three times, as previously. A 1:1 mixture of the kit’s color reagent A and B were mixed, then 200 µL was added to each well. After a 30 min incubation at ambient temperature, 50 µL of stop solution was added to each well. The absorbances were read on an M5e plate reader (Molecular Devices, San Jose, CA, USA) using a 450 nm detection wavelength and 540 nm as the blank wavelength for baseline correction. The quantified concentration of unbound VEGF in pg/mL was reported. All samples were analyzed in triplicate.
2.11. Solubility
To measure the equilibrium solubility of the API and excipients, a saturated solution of each substance was prepared by stirring the excess solids in the solvent for ~2 h. The solutions were centrifuged to remove excess solids. A total of 50 µL of the supernatant was pipetted into an aluminum pan and placed into a Thermogravimetric Analyzer (Discovery TGA, TA Instruments, New Castle, DE, USA). The sample was heated to 130 °C at a ramp rate of 50 °C/min and then held isothermal for 10 min to allow all of the solvent to evaporate. The final mass of the sample (of the known 50 µL volume) was recorded and used to determine the solubility in mg/mL. Samples were measured in triplicate.
3. Results
3.1. API and Excipient Solubility
Solubility screening was performed on the three small molecule APIs and the L-leucine excipient in water, 90/10 methanol/water and 80/20 ethanol/water by weight. These solvent ratios were chosen to maximize the solubility of L-leucine and API. Additional details on solubility and formulation selection can be found in the
Supplementary Information. As is evident in
Table 3, bevacizumab, erlotinib, and paclitaxel are not soluble in a common solvent at sufficient concentrations. This highlights the need for the simul-spray process to create a combination product. An aqueous buffer can be used for both bevacizumab and cisplatin, but it was found that bevacizumab and cisplatin were not chemically stable when co-dissolved in the same solution.
3.2. Residual Solvents
Water content is an important attribute of an inhalable formulation. If water content is too high, the recrystallization of amorphous domains, particularly trehalose, can cause failure. If the water content is near-zero, static can dominate the powder, resulting in poor release from the inhaler device. The water content of the simul-spray SDDs was quantified by Karl Fisher. Overall, all samples contained between 1 and 3.5% water by weight after manufacture.
While the water is retained and important to the formulation, residual organic solvent is undesirable from a safety perspective, and long-term exposure of the BEV to high levels of residual organic solvent could lead to degradation. With these spray conditions, the methanol (ERL-containing powders) and ethanol (PTX-containing powders) were both removed below the ICH limits of 0.3 and 0.5 wt %, respectively.
3.3. Drug Concentration
The API concentration in the simul-sprayed powders were quantified on a dry basis. The percent of theoretical concentration is shown for each active in
Figure 2. All of the small molecule drug concentrations are within 10% of target. The BEV concentrations are all slightly low, with the ERL 1:1 being 17% below target. A contributing factor to the low concentration of BEV is that proteins can adsorb to the surfaces of the liquid tubing (used to feed the spray dryer), which can lead to aggregation [
18]. In our previous work with a mono BEV formulation spray-dried with a similar configuration and formulation, 92.5% of target drug concentration was achieved [
17].
A likely cause of the drug concentration discrepancies is the water content differences between the formulations. When measuring drug concentration, the mass of water is subtracted from the total mass of the sample to provide a water-corrected concentration value. This is accomplished by using the bulk residual water content of the powder measured by KF. However, the water is retained differently by the individual formulations. For example, when equilibrated to the same ambient conditions, the weight percent of water for the BEV, ERL, PTX, and CP mono SDDs was 4.7, 0.6, 1.4, and 4.9, respectively. Equilibrated to the same conditions, the simul-sprayed powders show water contents close to what would be expected based on the formulation ratios, but still with some discrepancies, as shown in the
Supplementary Information. Therefore, the BEV values are likely to be slightly under-corrected, especially for the ERL and PTX-containing powders, which could contribute to the low concentration observed in most formulations. Similarly, the small molecule potencies might be slightly over-corrected.
3.4. Physical State of Simul-Sprayed Formulations
3.4.1. PXRD
Overall, DSC and PXRD confirmed that the physical state of the simul-spray formulations matches that of the mono formulations. X-ray diffraction was performed on the samples to determine the qualitative presence of crystalline material. In all formulations, characteristic peaks of crystalline L-leucine were found. The presence of crystalline L-leucine is an important formulation attribute for achieving good aerosol properties in spray-dried inhalation dry powders [
19]. PXRD confirmed that BEV, PTX, and CP are all amorphous within the instrument’s limit of detection. It also confirmed that ERL was present in crystalline form. A detailed discussion can be found in the
Supplementary Information.
3.4.2. DSC
Thermal analysis was conducted on the simul-spray formulations to demonstrate that the physical state of the materials was not altered by spraying in the presence of a second solvent. All simul-spray formulations showed glass transition temperatures characteristic of the BEV/trehalose phase, with an onset of ~118 °C. Additional glass transitions of the CP/trehalose phase (onset ~109 °C) or PTX phases (onset ~150 °C) were observed in corresponding formulations. Additional details of the thermal analysis can be found in the
Supplementary Information.
3.4.3. Morphology of Particles by SEM
BEV spray-dried formulations and small molecule spray-dried formulations have different surface features and morphology, making it possible to visually distinguish between the particles in SEM images. Exemplary SEM images of this is shown in
Figure 3, which depicts BEV, CP, ERL, and PTX mono-API formulations along with their corresponding 1:1 simul-spray formulations. The BEV particles have a smooth surface and are fairly collapsed, while the ERL, PTX, and CP particles have a corrugated surface and are generally closer to spherical in shape. The difference is most prominent for the ERL simul-sprays due to the presence of ERL as crystals.
The differences in surface appearance may be explained by competition for the receding interface of the drying droplet. It has been well-established in the literature that L-leucine-containing inhalation powders should be designed such that the L-leucine crystallizes out of solution during droplet drying, thus enriching on the surface (18). This maximizes its effectiveness as a dispersibility enhancer. It is likely that bevacizumab competes for surface enrichment due to its large molecular size. L-leucine crystal growth could also be delayed due to bevacizumab’s high viscosity near the droplet surface. These factors would lead to a smoother particle surface for the BEV particles as compared with the other APIs.
By visual inspection of the SEM images, it was clear that both components of the simul-sprays have approximately the same particle size distribution. This was then corroborated quantitatively by measurements of the aerodynamic particle size of the powders. Additionally, the images demonstrate the successful and independent atomization of the formulations. No fused particles were observed, confirming minimal to no interactions of the spray plumes before droplet solidification.
3.5. Aerosol Properties of Simul-Sprayed Formulations
The aerodynamic particle size distribution of the SDD powders was analyzed using an Aerodynamic Particle Sizer (APS) instrument with a powder disperser. For all formulations, the mass median aerodynamic diameter (MMAD) was between 1.5 and 3 microns, well within the target range of 1–5 microns for lung delivery. The results for MMAD and geometric standard deviation (GSD) are shown in
Table 4.
Full particle size distributions for three exemplary formulations are also shown in
Supplementary Information. Notably, the particle size distributions are monomodal in nature, which means that both types of particles within the simul-spray formulations have aerodynamic diameters that are in-range for pulmonary delivery.
A fast-scanning impactor (FSI) was used to quantify the fine particle dose (FPD) normalized by the capsule fill mass (FPD/fill mass). The FSI is similar to the more popular Next Generation Impactor or the Anderson Cascade Impactor, but a single stage is used to gravimetrically quantify powder with an aerodynamic diameter of <5 microns. A capsule containing 10 mg of the SDD powder was loaded into a dry powder inhaler for use with the impactor. A summary of the data for this test is shown in
Table 4. Overall, PTX simul-sprays had the highest %FPD/nominal, and ERL had the lowest.
To see if the formulations uniformly aerosolized, the FPD portion from the FSI testing of select samples were analyzed for drug concentration of the actives. Similar FPD/Fill mass values were found for both actives in each case, and more details are provided in the
Supplementary Information.
3.6. Anti-VEGF Activity of BEV Simul-Sprays
An ELISA-based binding assay was used to assess the ability of BEV in the simul-sprayed powders to disrupt VEGF binding. Briefly, reconstituted simul-spray powders were incubated with VEGF protein in the assay buffer. Next, the quantity of unbound VEGF in the solution was quantified by a commercially available VEGF ELISA assay kit. The lower the quantified level of VEGF from the assay, the more active BEV was at binding VEGF. As-received BEV stock solution was used as a positive control for the assay. Four negative controls were demonstrated: a solution containing VEGF and no actives, as well as VEGF/ERL, VEGF/PTX, and VEGF/CP solutions without BEV. These were used to demonstrate that the small molecules did not interfere with VEGF binding on their own.
Figure 4 shows the results of the VEGF assay. The positive control sample, using BEV stock, has 318 pg/mL of unbound VEGF. The negative control samples all read VEGF values of >3000 pg/mL, saturating the detection of the assay, as expected. The ERL, PTX, and CP negative controls all show that the small molecules do not inhibit VEGF measurably. Both the ERL 1:2 and 1:1 simul-sprays have similar values, which are the same as those for the positive control within the error of the assay. All PTX formulations show VEGF-inhibition levels similar to the positive control. All three cisplatin simul-sprays show VEGF inhibition. It is not known at this time why the CP 2:1 formulation appears to inhibit more strongly than the others, and detailed assay development is out of the scope of this work.
Overall, the activity assay confirmed that active BEV is present in the simul-sprays, with the preserved ability to inhibit VEGF after spray drying indicating its preserved biologic activity. This finding is similar to that of our previous work with spray-dried BEV monotherapies, where VEGF activity was quantified using a cell-based assay [
17].
5. Conclusions
This article demonstrated the simul-spray drying technology in which spray-dried particles of two different compositions are atomized into the dryer simultaneously with the use of two separate nozzles, forming a uniform powder blend. Simul-spraying was used in this work to manufacture combination therapies for inhalation which contain BEV and small-molecule cancer therapies: ERL, PTX, and CP. The resulting powders achieved their target drug concentration, good aerosol properties for delivery to the lung, and preserved anti-VEGF bioactivity. For lung cancer, inhaled combination therapies could locally treat this complex disease, easing patient compliance, reducing the dose, and limiting the exposure of healthy tissue to toxic compounds. More generally, a simul-spraying technique can be used to prepare combination inhalation therapies with otherwise incompatible APIs. The process could eliminate additional blending operations on poor-flowing inhalation powders and potentially shorten development timelines for combination inhalation products.