Nozzle Selection During Process Development of a Pharmaceutical Spray Dried Solid Dispersion

Transcription

1 Nozzle Selection During Process Development of a Pharmaceutical Spray Dried Solid Dispersion M.K.N. Hawk and K.A. Blakely * Small Molecule Design and Development Lilly Research Laboratories Indianapolis, IN USA Abstract Spray drying is a common unit operation across many industries due to the ability to control bulk powder properties and isolate dry product from a liquid feed in a single operation. For the pharmaceutical industry, spray drying offers the control of physical properties of crystalline or amorphous active pharmaceutical ingredients (APIs), as a neat or formulated drug product that can lead to improved bioavailability of the drug candidate. Drug candidate molecules are often formulated as a solid dispersion, incorporating both API and a polymer excipient for enhanced dissolution performance and improved physical stability. Bulk density and particle size of the spray dried dispersion can impact downstream manufacturability as well as performance of the dosage form. Robust ranges should be explored during development while minimizing the use of high-cost and limited supply API. For Compound A, formulated as a spray dried amorphous solid dispersion for oral delivery, atomization was decoupled from the drying process and a Phase Doppler Particle Analyzer was used to determine the appropriate nozzle for several different particle size targets. The selected nozzles were then used within a pilot scale spray dryer to produce product with the desired particle size and density combinations for downstream manufacturability testing. * Corresponding author: blakelyky@lilly.com

2 Introduction As the complexity of new pharmaceutically relevant compounds have increased, so has the challenge of efficiently and effectively delivering the active pharmaceutical ingredient (API). Estimates range from 40% and upwards as to the percentage of APIs currently in development that have limited aqueous solubility [1]. To address these challenges, the pharmaceutical industry has applied a variety of techniques to improve API properties for delivery, namely amorphous solids dispersions [2,3]. By preparing the active ingredient as an amorphous material, a higher energy state is attained, offering higher bioavailability. In order to promote physical stability of the higher energy amorphous form in the solid state, as well as inhibit crystallization during delivery of the active ingredient, the API is often stabilized within a carrier excipient (or inactive ingredient); typically a high glass transition polymer that is coprocessed with the API to form the intimately mixed solid dispersion. Spray drying is one of the most common techniques to produce an amorphous solid dispersion. Spray drying is used to 1) convert the API to the amorphous form by rapid solvent evaporation and 2) engineer the powder properties for optimal drug performance and successful downstream operation. Powder physical properties of interest are often density and particle size and may be identified as critical quality attributes of the spray dried product, reported and monitored with regulatory agencies, depending on the severity of impact to the patient and the ability to sufficiently control the process. Product density is controlled through the formulation of the solid dispersion (e.g., choice of polymer) as well as the selection of drying conditions within the spray dryer. Particle size is dictated by the size of the atomized droplet within the spray dryer. The selection of atomizer type is dependent upon the fluid properties. Based on the inclusion of a polymeric excipient in the amorphous solid dispersion feed solution a viscous feed solution is typical. Rationale for selection of atomizer type is well discussed elsewhere [4], leading to single fluid pressure nozzles as the atomizer of choice for most spray dried amorphous solid dispersions. The atomized droplet size is based upon the fluid properties of the feed solution and the atomization nozzle selected to produce the droplets. With each new investigational drug candidate, the properties of the molecule dictate not only the formulation of the solid dispersion, but also the selection of the feed solution solvent and operating conditions (e.g. temperature, pressure, solids concentration). Selection of the appropriate spray nozzle must account for an understanding of the fluid properties and droplet size targets, but also the thermodynamics of the spray drying operation. The nozzle operating pressure affects both the droplet size and the mass flow rate of the feed into the dryer for a single fluid pressure nozzle. Changes in atomization pressure impact droplet size, as well as the flow rate and hence the conditions within the drying chamber of the spray dryer. In order to decouple these processes, multiple nozzles are identified to 1) produce different product particle sizes at the same flow rate or 2) produce the same particle size across widely different flow rates, or 3) combinations thereof. Selection of a suitable single fluid atomizer for pilot scale spray drying operations will be discussed for Compound A, a poor soluble API current formulated as an amorphous solid dispersion prepared by spray drying. Objectives The objective of nozzle selection for Compound A was to sufficiently understand the atomization of the feed solution to meet the particle size target of 30 µm. This was accomplished in several stages: 1. Preliminary Nozzle Identification: determine likely nozzles to be used for the process, given nozzle characteristics and fluid properties 2. Water Flow Testing: quickly establish an acceptable stock of nozzles for further testing 3. Solution Characterization: to determine a suitable placebo solution 4. Pressure vs. Flow and Planar Spray Characteristics: Characterize a range of nozzles using solvent and placebo based solutions 5. Nozzle Recommendations: Select a nozzle(s) for pilot-scale studies Materials and Methods Preliminary Nozzle Identification In order to match the target particle size of the Compound A spray dried product, an understanding of the spray drying process parameters is important. The feed rate of the solution, not only contributes to the nozzle selection, but impacts dryer thermodynamics and operational efficiency. For pilot scale trials of Compound A, a maximum feed pressure of 1000 PSI and maximum feed solution flow ratesof 18 kg/hr were known based on pilot dryer equipment capability and capacity. The target powder particle size was based upon prior materials that had been used in clinical studies and the desire to match previous properties with a D v,50 = 30 µm. While no specifications exists for the full particle size distribution, downstream processing is aided by a narrow, monomodal particle size distribution. One approach proposed previously is to apply a mass balance around the droplet-to-powder formation process, relating the true density of the product with the solids concentration of the liquid [5]. In the case of polymer droplet drying, the polymer forms a skin at the surface of the droplet, which results in product particle sizes smaller than, equal to, or larger than the solution 2

3 droplet forming the particle dependent on drying conditions and product formulation. Currently, there is no reliable substitute for historical or empirical relationships, and often those determined or applied from the literature are for specific fluids or specific nozzle types [4,6]. Therefore, institutional knowledge along with experimental confirmation continues to drive nozzle selection. A range of commercially available nozzles (Table 1) were identified to meet the desired particle size based on historical experience. These include Type AF Mist Jet Hollow Cone nozzles (Steinen Manufacturing Company, Parsippany, NJ, USA), Schlick V121 nozzles (Düsen-Schlick GmbH, Untersiemau, Germany), and SprayDry SK nozzles (Spraying Systems Company, Wheaton, IL, USA). Selected nozzles produce a hollow cone pattern and include a range of orifice openings. Table 1: Preliminary Selection of Compound A Pilot Scale Nozzles, with manufacturer s operating ranges based on water Nozzle Make/Model Pressure Range (PSI) Flow Range (g/min) Steinen AF Steinen AF Schlick V Schlick V Spraying Systems SprayDry SKA Water Flow Checks Nozzle evaluation is conducted in several steps using a variety of feed equipment and instrumentation. The first step is a pressure versus flow assessment conducted using deionized water, while visually assessing the symmetry of the spray. Water testing is compared to published vendor data to see if similar nozzle performance would be expected across the same make and model of nozzle for future use (e.g., a like-for-like replacement). While this comparison does not confirm suitability of a nozzle across the operating range under expected feed conditions, it does provide a quick method to screen potential nozzles and for those that compare poorly, select a replacement, if desired, before further testing. Figure 1 depicts the nozzle test stand used during pressure flow testing and spray analysis. The nozzle test stand includes a stainless steel chamber with multiple sight glass positions and a manual 3-axis positioning traverse (Eli Lilly and Company, Indianapolis, IN USA). Feed equipment is interchangeable depending on the operation conditions. For this body of work the following feed equipment was used: a pump (HNP Mikrosysteme GmbH, Schwerin, Germany), pressure transmitters (Emerson Process Management Rosemount, Chanhassen, MN, USA), and a Coriolis mass flow meter (Emerson Process Management Boulder, CO, USA) monitoring temperature, density, and mass flow, all of which is monitored and archived by a data historian. For this experiment, a PowerSight Phase Doppler Particle Analyzer (TSI, Inc., Shoreview, MN, USA) with 2-axis traversing system, integrated with FlowSizer software, was used for droplet size and velocity measurements. Figure 1: Nozzle Test Stand for pressure versus flow testing and spray analysis by PDPA Solution Characterization A variety of different solutions were prepared and tested during the nozzle selection process, including: deionized water, 90/10 (w/w) acetone/water, and 4.04 wt% HPMC-AS MG (Shin-Etsu Chemical Company, Tokyo, Japan) polymer in 90/10 (w/w solute free) acetone/water. Solution properties were tested at time of use or had been previously tested by the Material Science and Physical Characterization lab at Eli Lilly and Company, including surface tension, viscosity, and density. Further discussion of these techniques are not included in this work. Pressure vs. Flow and Planar Spray Characteristics Nozzle testing using the 90/10 (w/w) solvent mixture without polymer (solvent) and the identified placebo solution were conducted on the nozzle test stand. Nozzle testing focused on establishing baseline pressure versus flow information for the identified solvent and placebo solution, followed by collecting planar spray characteristics of droplet size and velocity information at selected points along the nozzle flow curve. Both sets of information are often gathered in the same experiment for operational efficiency. The pressure versus flow information allows for comparison of likefor-like nozzles during current and future testing, as well as providing target operating points during start-up and shut-down of the spray dryer. Issues with a nozzle can often be detected prior to committing the feed solution (with API) to the dryer. Furthermore, observations 3

4 of the spray can be made during pressure versus flow testing to confirm the operating range of the nozzle that establishes a well formed spray (e.g. spray angle determination, image analysis, etc.) for the fluids being tested. As in the case of Compound A, fluids of interest often differ in properties from water, which is typically the fluid for the data available from nozzle manufacturers. Once an operating curve is established, conditions can be selected to further investigate planar spray characteristics of the nozzle. Several operating points were selected along the pressure curve to determine the droplet size vs. pressure relationship. PDPA is used to collect point measurements within and near the spray plume. The number and position of the point measurements varies depending on the familiarity of the system and the nozzle geometry. For hollow cone spray nozzles, symmetry can be used to reduce the number of analysis points to a single radii from the center axis of the spray to the edge of the spray. For the purposes of this work, the entire diameter of the spray was used for analysis. Several different analysis planes were selected at different distances away from the tip of the nozzle. Point measurement data collected from the PDPA system was then post-processed to understand planar spray droplet size as a single distribution. Applying a similar methodology to that previous reported by Bade and Schick [7], area and flux weighting were applied to determine droplet size statistics for a given analysis plane, at a given pressure condition for a given nozzle. By initially selecting a number of analysis planes below the nozzle, the impact of any droplet changes (e.g. evaporation due to volatile components) was desired to be minimized by selecting the most representative distances from the nozzle by comparability of the D v,50 statistic for adjacent analysis planes. Results and Discussion Water Flow Checks For Compound A, water flow checks quickly identified which nozzles would be suitable for further testing. Figure 2 illustrates an acceptable nozzle that compared well and an unacceptable nozzle, that needed to be replaced (like-for-like) and retested or not included in further testing. The SK nozzle was used directly for further testing, while nozzle ST did not have a replacement available and was not evaluated further. Figure 2: Water pressure versus flow testing comparing vendor data to experimental data for nozzle ST (poor comparability) and SK (acceptable) Placebo Solution Characterization For Compound A, the feed solution is a quaternary system, consisting of two solvents (acetone and deionized water), HPMC-AS polymer, and the active ingredient with an overall solids loading at 5 wt%. Rheological testing of the feed solution indicates the fluid is slightly shear-thinning. Prior testing has confirmed that the liquid surface is solvent saturated and the surface tension is dictated by the solvent as opposed to any of the solution additives. A suitable placebo (surrogate fluid) is also important to identify prior to nozzle testing, in order to 1) reduce (or eliminate) active ingredient consumption during nozzle testing and 2) reduce potential exposure to the active ingredient during handling and testing. Compound A, as in similar solid dispersion processes observed to date, does not contribute significantly to the rheology of the solution. As a result, similar rheology is obtained through a polymer only solution. A placebo solution for Compound A was prepared with various concentrations of HPMC-AS MG in 90/10 (w/w solute free) acetone/water. As shown for selected concentrations in Figure 3, the 4.04 wt% placebo profile matches well with the active solution profile. The viscosity for this feed solution is low (turbulent flow deviations were observed in this rheometer geometry at low concentrations, also evident in Figure 3) and exhibits near Newtonian behavior. This was not unexpected given the low solution concentration. 4

5 Figure 3: Selected placebo solutions (HPMC-AS MG at various concentrations) compared to active feed solution, and the effect of shear rate on viscosity Pressure vs. Flow and Planar Spray Characteristics The 90/10 (w/w) acetone/water solvent system in use for Compound A had not been investigated in the Small Molecule Design and Development (SMDD) spray analysis laboratory at Eli Lilly and Company. Given the new system and a desire to gain additional experience, additional solvent-only spray data was collected using the nozzle test stand apparatus to reduce the number and duration of the placebo spray tests. Using the information collected during pressure versus flow experiments, representative distances were selected to ensure droplet formation had occurred, but evaporation was kept to a minimum. Figure 4 illustrates data for the SprayDry SKA nozzle over a range of distances from the nozzle tip. As expected, the D v,10 did not change significantly with increasing distance from the nozzle, while the D v,90 changed rapidly close to the nozzle orifice, followed by a much slower rate of change at further distances from the nozzle. A distance of 20 mm was selected to represent this nozzle based upon the transition to slow decay of the D v,90 statistics and a stable D v,50 statistic. Figure 4: Droplet size as a function of distance from the nozzle for SprayDry SKA nozzle operating at 62 PSIg with 90/10 (w/w) acetone/water The pressure versus flow curve aids in the selection of operating conditions for droplet size analysis. Selecting multiple points along the pressure versus flow curve determines the droplet size over a range of potential operating pressures for a given nozzle. As illustrated in Figure 5 for 90/10 (w/w) acetone/water for a Steinen AF nozzle at two different operating points at a 20 mm distance from the nozzle, the droplet size is moderately affected by a change in operating pressure. Figure 5: Solvent mass flow rate as a function of pressure, with measured droplet size at selected pressures for Steinen AF nozzle Building upon the added experience of the solvent only testing, the placebo testing was implemented more rapidly through a reduced number of analysis planes and an overall reduction in the number of point measurements. Placebo solution droplet size data was suc- 5

6 cessfully obtained for only one of the nozzles of interest prior to the start-up of the pilot scale spray drying campaign. Based on solvent flow curves and past experience, the SC nozzle was selected for further evaluation. Figure 6 illustrates droplets sizes were not appreciably different between solvent-only and 4.04 wt% HPMC-AS MG placebo feeds atomized by the SC nozzle, suggesting solvent only testing might be sufficient for future droplet size testing. However, even solvent only droplet testing can be used in the future, both fluids will still need to be tested to establish their respective pressure versus flow curves. Figure 7: Mass flow for water, solvent, and placebo solution as a function of operating pressure, with measured droplet size for solvent and placebo solution feeds at selected pressures for nozzle SC Figure 6: Mass flow for water, solvent, and placebo solution as a function of operating pressure, with measured droplet size for solvent and placebo solution feeds at selected pressures for nozzle SC SC was used to produce placebo product on the pilot scale spray dryer with somewhat surprising results. The placebo powder D v,50 particle size post spray drying over a variety of nozzle pressure set-points is illustrated in Figure 7. Little to no change was observed in the placebo product particle size, even with expected changes in the droplet size due to different atomization conditions. This would suggest that the drying conditions within the spray dryer may control product particle size more than minor changes in atomization conditions. The SC nozzle produced material at the D v,50 target of 30 µm. Due to project constraints, placebo droplet size testing was not completed for nozzle ST prior to testing on the pilot scale spray dryer. ST was included in the placebo spray drying campaign to evaluate atomization from a slightly higher pressure drop/smaller orifice nozzle. As expected, slightly smaller product particles were produced (Figure 8). Figure 8: Mass flow for water, solvent, and placebo solution as a function of operating pressure, with measured droplet size for solvent and placebo solution feeds at selected pressures for nozzle ST Combining historical experience, pressure versus flow testing, and PDPA droplet size data, several different hollow cone nozzles were evaluated to produce Compound A amorphous solid dispersion with a D v,50 = 30 µm. Hollow cone nozzles were selected and tested against manufacturer specifications, prior to further testing with solvent and placebo solutions. The placebo solution composition was confirmed by comparison of solution rheology. Ultimately, two nozzles were selected for use in the pilot scale spray dryer, producing placebo at the D v,50 = 30 µm target (SC ) and slightly smaller size particles (SC ). Nozzles were subsequently used for further placebo and active feed solution spray drying with successful demonstration of product with on-target particle size. 6

7 Acknowledgment The authors wish to acknowledge the help of the following persons in technical discussions and support: Dr. Christopher Burcham, Dr. Paul Stroud, Mr. Douglas McKinney, Mr. Joshua Schenck, Dr. Dushyant Shekhawat, and Dr. Steven Doherty. References 1. Shah, N. ed, Amorphous Solids Dispersions: Theory and Practice, Springer, 2015, p Merisko-Liversidge, E., Liversidge, G.G., Cooper, E.R., Eur. J. Pharm. Sci. 18: (2003). 3. Newman, A., Pharmaceutical Amorphous Solid Dispersions, Wiley, 2015, p Shah, N. ed, Amorphous Solids Dispersions: Theory and Practice, Springer, 2015, p Maa, Y.-F., Nguyen, P.-A., Sit, K. and Hsu, C. C. (1998), Spray-drying performance of a bench-top spray dryer for protein aerosol powder preparation. Biotechnol. Bioeng., 60: doi: /(sici) ( )60:3<301::aid-bit5>3.0.co;2-l. 6. Lipp, C.W., Practical Spray Technology: Fundamentals and Practice, Lake Innovation, 2012, p Bade, K.M. and Schick, R.J., ILASS-Americas 25th Annual Conference on Liquid Atomization and Spray Systems, Pittsburgh, Pennsylvania, May