(583a) Impact of Rheology and Process Variables on the Development and Scale-up of a Drum Filling Process for Dry Powder Inhalers

Dry powder inhalers (DPI) are becoming increasingly popular to treat both respiratory and systemic diseases as they provide better physical and chemical stability of the drug product. Traditional DPI formulations have been composed by physical mixtures of coarse carriers with one (or more) micronized active pharmaceutical ingredients (API), typically with aerodynamic particle sizes below 5 µm. While being a well stablished formulation strategy, carrier-based drug products can suffer from lack of homogeneity, inefficient drug deposition and are limited to APIs with good solubility.

Composite particles are therefore receiving increased attention as an alternative formulation strategy. In these formulations, the API is embedded in an excipient matrix, in single engineered particles, typically with low densities and small aerodynamic particle sizes, that can be efficiently delivered to the lung for improved performance. However, these unique properties can also result in operational challenges while processing these powders. Due to their large surface area and low densities, composite powders can present a highly cohesive behaviour, ultimately resulting in poor flowability and intense aggregation during manufacturing steps as capsule or device filling. Moreover, DPI formulations may contain doses as low as a few milligrams, meaning that any minor deviation resulting from the flowability issues can jeopardize the feasibility of the manufacturing process.

In DPIs, capsules are a widely used solution to work as depots of the formulation to be later loaded into a device and actuated by the patient for administration. To fill these capsules, several manufacturing systems are currently commercially available, although not all of them are completely suited for low-dosage filling of composite formulations. On that domain, vacuum drum filling has been gaining traction as its working principles are thought to handle poorly flowable powders more accurately without compromising the scalability of the process. Indeed, drum filling has been capable of achieving target fill weights down to 1 mg which can be considered challenging with other high-speed filling technologies. However, despite the increasing interest, few studies can be found addressing process development and scale-up of this technology and, most importantly, how powder properties of different inhalation formulation can impact its performance.

The characterization of the rheological properties of inhalation-based powders can provide valuable in-sight for both formulation and process optimization as one tries to achieve a specific fill-weight and aerodynamic performance in the capsule/device. For dispensing and filling purposes, especially for low dosages, powder compressibility, density, cohesiveness and ultimately flowability can be derived from analytical characterization techniques, such as Shear Cell. However, limited information is available regarding the relation between these types of rheological properties and the operational space of vacuum drum filling processes.

The goal of this study is then to assess the impact of formulation and process variables in the performance of a capsule filling process using vacuum drum filling. First, two representative powders of DPI formulations were selected to assess the impact of rheology in the operation conditions of a drum filling process. A composite spray dried powder of 80:20 % w/w trehalose dihydrate and L-leucine and a micronized crystalline API-alone (Compound A) formulation were selected as case-studies for this work. Particles with d(0.5) ranging from 2.2 up to 2.8 µm were obtained for the composite formulation as a function of spray drying parameters, while for Compound A micronization conditions were adjusted toward achieving a d(0.5) of 2.8 µm, without any further processing. The fundamental rheological properties of both materials were determined using the yield locus test under a pre-shear normal stress of 2 kPa and conducted at three increasing normal stresses.

Then, HPMC size #3 capsules were automatically filled using lab-scale Drum Lab (Harro Höfliger) at a target weight of 10 and 20 mg using a 7.5 mm³ dosing bore volume at the Drum Lab. For each test 100 capsules were filled, while the powder level on the feeding hopper was maintained constant manually. For the homogenization of the powder bed, a wire-type stirrer was used. Preliminary studies allowed to define the target vacuum pressure to assure the target fill weight, then an operational range was defined until the edge of failure in terms of fill weight was found. The process performance was evaluated as a function of the obtained relative standard deviation (RSD) for central point and process acceptable ranges limits (PAR). Both formulations were sieved using a 600 µm mesh before feeding the hooper. Additionally, to test the impact of scale in process variables at different scales, the same capsule filling methodology was performed a pilot scale ModuC-LS (Harro Höfliger), with the composite formulation. At this scale, capsules were filled at a target weight of 8 mg using a 15 mm³ dosing bore volume while for the scale-up tests a target weight of 5 mg using a 7.5 mm³ dosing bore volume. In the pilot scale unit, powder level on the feeding hopper was automatically controlled and kept constant. Finally, filled capsules were characterized for relative water content (by KF-oven) and aerodynamic performance by Anderson Cascade Impactor (ACI).

Lab-scale results showed that filling parameters, namely vacuum pressure, is highly dependent on powder rheology and cohesiveness. Both formulations had different flow properties, with a flow function (F) of 1.3 and 2.3, for the composite and API-alone formulations, respectively. These measurements are indicators that the API-alone formulation is less cohesive than its composite counterpart, as expected. This prediction was later confirmed during the capsule filling process as, for the same target fill weight, higher vacuum pressure was required for the API-alone formulation (-110 mbar, comparing to -50 mbar for the composite). Furthermore, the latter also had a tighter vacuum pressure range (10 mbar) to achieve the same target 10 mg fill weight. At the same vacuum pressure, achieving 20 mg for the composite formulation was possible just by increasing the number of dosages plugs. Despite the differences in both formulations, the relative standard deviation (RSD) of the filling weight equally increased with lower vacuum pressures. Such increase can be attributed to less compacted plugs that can eventually result in higher fill weight variability. On the other hand, too high vacuum pressures can potentially impact the performance of the DPI, as the resultant compacted material may prevent the redispersion of the plug and hinder aerosolization and efficient delivery to the target pulmonary stages. Results from aerodynamic performance resultant from each vacuum pressure should be performed to each specific formulation to complement the acceptable range of filling weight variability, and together generate the design space of the capsule filling process.

For the scale-up evaluation, results confirmed similar response of filling performance to differences in vacuum pressure at both lab and pilot scale units. However, it can be concluded that lab-scale trials always performed worse than the ones performed in the pilot unit, i.e. higher RSD is always obtained in the former, for the same process conditions. Nonetheless, both set of results indicate that lab-scale trials can be used for process development and de-risking tool of the process at scale as: (1) qualitatively one can obtain the same response to process variables at pilot scale and (2) the process developed at lab-scale will always be a worst-case from the manufacturing scale. Additionally, it was found that sieving plays an important role in reducing the RSD of the capsule filling process of the composite formulation. Its known hygroscopicity and high surface area can promote the formation of aggregates that can be detrimental for the capsule filling process. A sieving step can segregate these aggregates and reduce powder static energy and, consequently, improve the flowability of the powder. This can also be beneficial for the performance of the DPI, as indicated by the higher emitted doses obtained using sieved material 90.3% comparing to 76.4%. Finally, stirrer offset was found to not significantly impact powder fluidization nor capsule filling performance.

Overall, the results from this work provide valuable in-sights for the development of a suited drum filling process from very early stages. Despite there is a clear dependence of the operational ranges of the process on the type of formulation, the outputs of a rheological characterization can be correlated with recommended process parameters values or ranges, even before performing any trial. Moreover, lab-scale trials have proved to be suited for developing and de-risking the capsule filling process at scale by representing a worst-case for the latter, while having similar sensitivity to the main process parameters (e.g. vacuum pressure). Additional work is still required to extend some of these conclusions to a wider range of rheological powder properties, but these initial results show great potential for developing improved process development methodologies for capsule filling processes at both lab and manufacturing scales.