Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Issues Associated with the Manufacture of Large Scale Crystalline Nanosuspensions

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Small volume crystalline drug nanosuspensions manufactured at AP using Fritsch P7 Planetary Micromill with zirconia milling pots & zirconia beads. Fritsch P7 Mill limited to relatively small volume manufacture. ~ 32ml Dynomill was acquired to manufacture large volume crystalline nanosuspensions –used stainless steel mill chamber and high density/highly cross-linked polystyrene milling beads (zirconia chambers available but very expensive) Can process litres of suspension Suppliers of polystyrene milling beads DOW (had an exclusivity agreement with ELAN). AZ obtained beads from NORSTONE. Mid 2010 NORSTONE signed exclusivity agreement with “?” – would not supply beads to AZ in future – alternative bead required. History Manufacture of crystalline nanosuspensions at Alderley Park

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY History A feasibility study was carried out at Sodertalje to identify an alternative. Trials used glass beads as it was perceived there might be issues with zirconia (heavy metal) for clinical manufactures. Sodertalje investigated soda lime & borosilicate beads of different sizes from suppliers such as:- Sigmund-Lindner, Sigma, Mo-Sci, Retsch-VWR & Christian Berner.

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Sigmund-Lindner Sigma Mo-Sci Retsch Glass Beads Before and After Milling Run in Dynomill Christian Berner

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Manuf./Supplier Type 1 Size [µm ] pH bef/aft Milling efficiency Milling observations Abrasion (SEM) Metals [AU] (EDS) Residuals [weight%] AZ PS / 6.1 Sigmund Lindner SL / 11.5 Sigmund Lindner SL / 11.0 Sigmund Lindner BS / 10.3 Sigma SL / 10.8 Sigma SL / 10.4 MO-SCI BS / 7.6 MO-SCI SL / 11.5 Retsch/VWR SL / 11.5 Retsch/VWR SL / 11.5 Christian Berner SL / 11.2 Alternative Beads Selected Based on Performance versus Polystyrene Beads Much worse than control Better than or equal to control Slightly worse than control

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Crystalline Drug Nanosuspension Manufacture Initially a nanosuspension was prepared by bead milling (using Fritsch P7 Planetary Mill & zirconia milling media) as limited exposure had been observed with a microsuspension formulation. M.E. Commonly use 3 prototype drug formulations dependant on the physical properties of the drug Initial nanosuspension was manufactured at 200mg/ml drug load in vehicle PVP/Aerosol OT and had mean particle size <200nm.

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Structures of Stabilizing Surfactants Dioctyl sodium sulfosuccinate (Aerosol OT ex Cytec ) (anionic) Polyvinylpyrrolidone ( ex BASF ) (non-ionic) DPPE-PEG2000 (Dipalmitoylphosphatidylethanolamine - poly(ethylene glycol) 2000 ( ex Genzyme ) (anionic) Hydroxypropylmethylcellulose ( ex Dow ) (non-ionic) Polyoxyethylene (20) sorbitan monooleate (Tween 80) (non-ionic)

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Crystalline Drug Nanosuspension Manufacture MTD and DRF studies required the manufacture of several litres of nanosuspension – Dynomill manufacture. Bead milled drug in Dynomill at 200mg/ml using vehicle PVP/Aot with MO-SCI soda-lime glass beads

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Crystalline Drug Nanosuspension Manufacture Measured Levels in suspension: Limits by EMEA: Fe: 42 ppm 1300 ppm Ni: 7 ppm 25 ppm Cr: 10 ppm 25 ppm Knew from feasibility study - Soda-lime glass beads would cause suspension pH to rise > pH9, and also metal and colloidal glass contamination. Measured pH of suspension was ~pH9.1 Analysis showed the metal levels were below acceptable limits.

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Crystalline Drug Nanosuspension Manufacture pH Adjustment: Attempted to manually adjust pH to pH6.9 by addition of 1M HCl – but reduced pH to ~pH2 which resulted in coagulation of the nanosuspension A nano suspension at pH 6.9 (from Fritsch mill) was manually pH adjusted to pH 9.1 and then successfully re- adjusted to pH 6.9 using 0.1 M acetic acid (weak acid). This nanosuspension remained as free flowing particles no visible evidence of aggregation However after addition of HCl adjusting pH to ~ 2, the nanosuspension coagulated. What would happen when dosed orally? Nanosuspension also coagulated when diluted in Simulated Gastric Fluid

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Crystalline Drug Nanosuspension Manufacture Preparation of nanosuspension at 200mg/ml in PVP/Aot in Acetate Buffer at pH4.6 – was successful d10 74nm D50 168nm d90 503nm

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Preparation of nanosuspension in DPPE-PEG mg/ml A 200mg/ml suspension in 4% DPPE-PEG2000 milled on the Fritsch mill produced a suspension with a similar overall size profile to that obtained with PVP/Aot: d50= 144nm d90= 644nm When diluted into SGF the DPPE-PEG2000 nanosuspension DID NOT COAGULATE but remained as free-flowing individual particles. However problems as to the suitability of this vehicle due to possibility of anaphylactic shock in dogs (due to peroxide content) also doubts about whether DPPE-PEG2000 can be taken forward into FTiM studies. Crystalline Drug Nanosuspension Manufacture d10 69nm d50 144nm d90 644nm Formulation was therefore deemed not suitable for use in the dog tox study and study postponed. Mouse study went ahead as only required small volumes which could be manufactured using the Fritsch mill Dog PK study was set up to assess whether there was any advantage in dosing a nanosuspension compared to a microsuspension Although there was variability in exposure, the results indicated the dog MTD could commence with a microsuspension, whilst further work progressed on the nanosuspension formulation However tox observed with microsuspension.

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Flocculation/Coagulation Testing of Drug Microsuspensions in PVP/Aot Microsuspension (200mg/ml) in PVP/Aot Diluted 1:3 in SGF adjusted to pH4.5 Diluted 1:3 in SGF (pH1.2) Carried out Flocculation/Coagulation testing of microsuspension formulations Diluted microsuspension manufactured in PVP/Aot and manufactured in HPMC/Tween80 in SGF and in SGF adjusted to pH4.5 (mimic mouse) √ = no coagulation X = coagulation PVP/Aot is not a good stabiliser system for this compound VehicleDilute in SGFDilute in SGF/pH4.5 HPMC/Tween80√√ PVP/AotX√

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY SUMMARY Nanosuspension manufactures using SL glass beads will result in high pH suspensions unless buffered PVP/Aot The PVP/Aot vehicle makes suspensions at 200mg/ml in the Fritsch Mill using zirconia media that are chemically and physically stable. The suspensions in PVP/Aot coagulate spontaneously when diluted into SGF (1:3 dilution) or when pH reduced to ~pH2 A nanosuspension can be prepared at 200mg/ml in PVP/Aot/acetate buffer(pH4.6) – but the suspension coagulated when diluted into SGF or pH reduced to ~2 DPPE-PEG2000 A nanosuspension could be prepared at 200mg/ml drug load in 4% DPPE-PEG2000 which does not coagulate when diluted in SGF Issues found with milling with glass beads – pH rise, some glass and metal contamination (albeit below recommended limits). Could be an issue for compounds unstable at high pH – but it is possible to manufacture in vehicle/buffer Accidentally found out that the nanosuspension in PVP/Aot coagulated at gastric pH. Also found out microsuspension in PVP/Aot coagulated at gastric pH indicating that PVP/Aot not a good stabilizer for this compound

Proprietary and Confidential © AstraZeneca 2009 FOR INTERNAL USE ONLY Recommendations M.E. do not routinely carry out flocculation/coagulation testing on suspensions – maybe we should? If it had not accidentally been found that the nanosuspension coagulated at gastric pH the nanosuspension would have been dosed, probably would have coagulated in the stomach, producing low exposures and variable results Future reference:- Important to choose optimum stabiliser system for nanosuspension manufacture