(88g) Manufacturing Drug Substance That Meets Formulation Needs

It is widely acknowledged that small organic molecules employed as pharmaceuticals can show variation with respect to their solid-state properties. The final drug substance particles are normally produced by of a series of unit operations executed in batch; typically crystallization, filtration, drying and, if necessary, particle size reduction. So considering the number of unit operations that the solids experience and the batch wise nature of their manufacture, it is little wonder that batch-to-batch variation is often observed. Thus, over the last few years there have been a number of programs conducted within AstraZeneca to better understand and control the manufacture our drug substances (DS). The main focus of the program was on control of particle size, shape and form through seeded crystallization strategies. In addition, we have generated a better understanding the impact of filtration and drying of those particles. This work has provided more consistent particles that meets the needs of the formulation teams. In this paper we describe AstraZeneca's ?Particles Size Control? program.

.

? Seeded Crystallization to produce more consistent crystals of varying size and shape.

It was found that it was possible to control the particle size of crystals in a cooling crystallization by using appropriate seeding conditions and keeping supersaturation low to avoid secondary nucleation. By varying the seed size it was possible to produce material with median sizes between 10 and 40µm, avoiding the bimodal particle size distributions seen in standard crystallisations, and to do so reproducibly. Materials characterisation showed these batches had favourable compression properties compared to the control, even after particle size reduction.

? Supersaturation Feedback Contol using ATR UV/Vis Spectroscopy

By maintaining a constant, low supersaturation during crystallization encourages crystal growth in a seeded cooling crystallization. ATR UV/Vis spectrosopy has been used not only to monitor crystallizations, but also, when properly calibrated to monitor and maintain constant supersaturation using a temperature feedback loop. To date, this has been shown be effective in laboratory investigations to produce significantly better crystals (larger, less agglomerated) compared to an unseeded cooling crystallization. This has now also been demonstrated to work at pilot scale (125L) and investigations at the 1000L scale are in hand.

? Wet milling of APIs to improve their bulk density and flow properties and provide material of varying grades for formulation optimization.

Wet milling provides an alternative to dry milling, since controlled particle size reduction is achieved by using an in-line rotor-stator device to provide high shear to a crystal slurry. Because wet milling can be easily incorporated on-line during or after crystallisation, it obviates the need for a separate dry milling step. To achieve a target particle size profile, wet milling can be used either alone or in conjunction with other techniques, such as temperature cycling of the slurry, to further improve the crystal size distribution. This technique has been used on a number of AZ compounds to improve their bulk density and hence flow properties.

? Temperature cycling Temperature cycling has been used to modify the habit of needle-shaped crystals and hence mitigate against problems associated with needles, such as poor filtration and low bulk density.

? The use of a continuous oscillatory baffled crystallizer (COBC) to enhance the consistency of materials and accelerate their manufacture.

Due to the increased specific area per volume for heat transfer together with the plug flow conditions in the COBC, it was demonstrated that continuous crystallization using a COBC overcomes many of the problems faced in traditional batch operations. Since the scale up of the COBC is linear and involves very small changes in diameter, this ensures that the governing science of solution crystallization remains the same when scaled up. This is a important factor that cannot be achieved in scaling up of a batch stirred tank batch system. The combination of these features showed that the COBC was able to deliver the isolation of the model API in just over 12 minutes compared to the 9 hours and 40 mins compared to a batch process.

? Understanding of the effect of filtration pressure on particles and measurements to predict their propensity for breakage during this operation.

The particle size distribution of crystallised material was compared to the particle size distribution of filtered material and dried material. It was found that for AZ compounds as well as commercially available materials, at lab and plant scale that the majority of the breakage was observed during pressure filtration, and to a smaller extent during the drying process.

The observed behaviour could be explained by describing the stress particles are subjected to when a bed is under compression. At typical conditions the stress on a single particle was found to be in excess of the tensile strength of typical pharmaceutical materials. As a result, the particles break reducing the particle size distribution. In some cases the size reduction is catastrophic, and in general, the bigger and more anisotropic the particles, the more likely breakage will occur.

To predict which particles are more likely to break, a technique was developed using dry dispersion laser diffraction particle sizing (Sympatec Helos/Rodos/Aspiros). A correlation was observed between (i) the fines produced on increasing dispersion pressure and (ii) breakage observed on pressure filtration. This provides a predictive tool to identify fragile particles, and thus the likelihood that the PSD will change due to pressure filtration. This will than forewarn formulators of the expected PSD of the API produced in the plant.