Challenges in Scaling Up Newly Developed Microbial Manufacturing Processes

Transcription

1 Challenges in Scaling Up Newly Developed Microbial Manufacturing Processes Susan Dana Jones, Ph.D. BioProcess Technology Consultants BIO Process Zone Theater June 19, 2008

2 Microbial Process Development Microbial manufacturing processes are developed individually Process is dependent on the unique structure and behavior of the product Platform processes are usually not applicable and therefore the development timeline can be longer. Scale up of a microbial manufacturing process requires significant process evaluation and rigorous analysis of multiple parameters Normally initial scale up is from 5-10 L lab scale to intermediate scale of L Secondary scale up from 100 L to 1000 L driven by data from initial scale up Two step scale up to production scale mitigates risk but is time consuming With sufficient information, scale-up directly from lab scale to production scale is feasible This approach will save time and money for companies that are developing unique products expressed in microbial systems

3 Example Scale up from 5 L to 1000 L Process was developed at the 5 L scale and parameters were fixed based on multiple process demonstration runs at that scale. Scale up to the 1000 L scale was seamless in most parameters due to the wealth of information obtained at the small scale. Some unit operations did not transfer perfectly Rigorous development and testing at small scale can enable transfer directly to large scale manufacturing

4 Project Overview Production host is an E. Coli K12 derivative Product is ~25kD recombinant protein expressed intracellularly Inclusion bodies are not formed at current expression levels Product is soluble intracellular Inducible expression using T7 promoter with IPTG induction of T7 polymerase expression Product is a vaccine candidate in early development Minimal process development to date Some analytical methods in place

5 Project Overview Shake flask fermentation process in place Low expression levels Complex medium, animal origin components No downstream process defined Research cell bank available Master cell bank required Process development initiated with the RCB Product not well characterized Robust analytical methods required Analytical support is vital to making good choices during process development

6 Customer Requirements Bulk clinical material in 12 months Manufactured at scale (1000 L) Minimal requirements 2 g purified and meeting specifications Process development focus is to quickly generate material meeting specs No time for full process optimization Process must be scalable Process must yield sufficient material at scale

7 Process Development To cut time, use established standards (medium, conditions etc) for E.coli fermentation No animal derived media components No complex feeds or supplements other than ph and do2 control Knowledge of facility operations at large scale influenced process development choices at 5 L scale Simple batch operation (quick, robust) Minimize downstream steps while achieving product quality and purity Use half of fermentation culture for downstream processing at scale

8 Analytical Development Analytical methods are key to successful process development Important to understand what the target product attributes will be Purity Size and structure Potency Methods to measure all these attributes must exist before commencing process development Interim reference material must exist Material from innovator s laboratory runs is sufficient for initial reference standard

9 Analytical Methods in Process Development Assay SDS page Western blot RP-HPLC Application Quick screening of effects on product expression with different process parameters, Relatively easy indication of impurity profile Poor release assay, subjective and not quantitative Non subjective measurement of purity and concentration Ideally suited to in-process control (rapid, accurate) Excellent release criteria Total protein DNA Endotoxin Used to track downstream processing development Relationship A 280 vs total protein to establish in process measurement using A 280 Commercial kit Commercial kits available. Establish spike and recovery

10 Analytical Challenges with Product Assay Total protein DNA RP-HPLC Challenge Product could not be measured using BCA protein assay due to shortage of key amino acids Protein interference in assay was resolved by DNA extraction Existing method took too long to be useful. Method time cut from 40 mins 6 mins with modification of gradient. Speed required for IPC

11 5 L Process Description Upstream Downstream Pre-culture (shake flask) Fermentation Cell disruption Batch centrifuge Depth Filtration ciex chromatography Ultra-filtration 8 kd Gel Filtration (Fractionation) 0.22µm Filtration Ultra-filtration 10kD Bulk fill

12 Scale up Strategy and Considerations Verify all operating parameters at 5 L scale with 3 verification batches prior to engineering batch at scale Use raw material lots intended at scale Linear scale up for chromatography and filtration steps Collect fractions in chromatography steps to allow for flexible pooling strategy at scale (good IPC required) Break through studies at small scale can support choice of filter area at scale Commercial availability and sizes of some units determines choice at scale Blocked filters can be easily replaced (design in) at scale but may impact overall process yield Process hygiene generally improves at scale IE, endotoxin levels always decrease

13 Scale up Considerations Process time increases with scale- product stability can be an issue 5 L process indicated homogenate was stable for 12 hrs at 4 o C Process volumes need to be manageable volumes <250 L are mobile and generally easy to cool/store Increasing column size gives increased pressure at scale and possible lower linear flow rates

14 Column Scale-up and Sizing Balance Cost of Production (COP), Production Rate (g/hr) and Productivity (g/hr/l support) Trade-off between capacity, throughput, and cost Multiple cycles vs. a larger column? Wider or taller? How frequently should media be replaced? Support equipment sizing and cost

15 Column Operating Options Cycling Small Column in Series Multiple Small Columns in parallel Single Large Column

16 Scale up -Volume and Time Step 5L scale 1000L scale Volume Process time Volume Process time Pre-culture I/II 1 x 150ml hrs 1 x 150ml, 4 x 2.3L hrs stage hrs stage 2 Fermentation 4.5L 9-11 hrs hrs 4 o C 1000L 9-11 hrs hrs 4 o C Cell disruption 4.5L 0.5 hrs 500L 1.5 hrs Centrifugation 4.5L 1 g (batch) 500L x g (continuous) Depth Filtration 4.5L 2 hrs 500L 6 hrs Ultrafiltration 0.75L 5 hrs 120L 3-5 hrs Cation Exchange 750ml (load), 180ml (eluate) 2 hrs 0.3 hrs 120L (load), 10L (eluate) 2 hrs 0.3 hrs Ultrafiltration 30ml 2 hrs 1L 2 hrs Gel Filtration 3ml fractions 3 hrs 250ml fractions 3hrs

17 Scale up - numbers Unit operation 5L scale 1000L scale Centrifugation Batch centrifuge Continuous centrifuge Depth Filtration Cuno m 2 /L harvest Cuno m 2 /Lharvest Cuno m 2 /L harvest Cuno m 2 /L harvest Lifeassure 0.014m 2 /L harvest Lifeassure 0.014m 2 /L harvest Ultrafiltration Cation Exchange Ultrafiltration Gel Filtration Final Filtration 0.1m 2 (0.01m 2 /L harvest) 2.6cm diameter 5.3 cm 2 surface area 15 cm bed height 0.05m 2 (0.27m 2 /L harvest) 1.6cm diameter 2.0 cm 2 surface area 67 cm bed height m 2 5.0m 2 (0.01m 2 /L harvest) 30 cm diameter cm 2 surface area 15 cm bed height 0.2m 2 (0.2m 2 /L harvest) 14 cm diameter cm 2 surface area 67 cm bed height 0.01m 2

18 Process Reproducibility at 5L Scale Variation in fermentation at 5L scale Optoical density 600nm Time (hrs)

19 Fermentation- 5 L vs L 30 Optical Density 600nm Time (Hrs) 1000L GMP 1000L Engineering 5L development

20 Fermentation In Process Controls Optical density ph Temperature Dissolved Oxygen concentration

21 Cell Disruption- 5 L vs. 500 L % reduction OD600nm L 5L Number cycles 1 In Process Controls: Optical density reduction >95% initial value Temperature

22 Centrifugation Difficult to scale batch continuous centrifuge Sedimentation models provide indicators for initial parameters Fixed speed and adjust flow to increase/decrease retention time Particle distribution pattern changes Often see product loss

23 Depth Filtration at Full Scale Life Assure filters blocked during first batch at full scale (Engineering Batch). Flow rate decreased in GMP batch to prevent similar blockage Centrifugation and filtration are challenging to scale up and are not a direct process transfer from development to production. Parameter Flow rate (L/hr) CUNO 60 (m 2 ) CUNO 90 (m 2 ) Life Assure (m 2 ) Engineering GMP

24 ciex Chromatography 91 ml Vs L

25 Ultra filtration At small scale dead volume of system, availability of material can limit ability to reach desired conc. Typically reach 50-70% of the desired product concentration at small scale Use Eng. Batch to verify the concentration step This approach works because the unit operation is reliable and scaleable. In interest of reducing process times excess filter surface area chosen for scale allows higher flow rates.

26 Gel Filtration Small scale (135ml) vs. Large scale 10 L

27 Gel Filtration Pooling of Fractions based on Purity and concentration purity (%) mg/ml Fraction A 60 0,01 Fraction B 50 0,04 Fraction C 72 0,28 Fraction D 92 2,78 Fraction E 94 3,74 Fraction F 95 1,84 Fraction G 92 0,69 Fraction H 86 0,38 Fraction I 85 0,23 Fraction J 76 0,06

28 Summary Unit Operation Pre-culture I/II Fermentation Centrifugation Cell disruption Depth Filtration Scaleabilit y Good Good Poor Excellent Good/ Excellent Comments Ratio of medium to flask volume is an issue for OTR. Additional pre culture phase for at scale applications needs to be tested during verification runs Process simplicity is the key with batch processes being most reproducible. Comparison of batch centrifuge vs continuous centrifuge remains challenge. Highly reproducible. Defining reduction in optical density allows differences at scale to be compensated Scale up on basis of surface area usually means larger margins at scale but are acceptable if load studies not executed.

29 Unit Operation Ultra filtration Chromatogra phy Scaleability Good/ Excellent Good/ Excellent Comments Surface area and trans membrane pressure should be monitored Establish binding capacity (for binding steps) early and test expected final bed height during development phase. Consider volumes of pools at scale when developing model collection and storage issues Consider flow rate effects at scale leading to pressure

30 Acknowledgements SynCo BioPartners Paul Ives Joanne McCudden APD Team BioProcess Technology Consultants Howard Levine James Blackwell

31 THANK YOU! BioProcess Technology Consultants, Inc. 289 Great Road, Suite 303 Acton, MA USA (phone) (fax)