UCL Decisional Tools Research Operational Economic Evaluation

UCL Decisional Tools Research Operational & Economic Evaluation of Integrated Continuous Biomanufacturing Strategies for Clinical & Commercial m. Ab Production Suzanne Farid Ph. D CEng FIChem. E Reader (Associate Professor) Co-Director EPSRC Centre for Innovative Manufacturing UCL Biochemical Engineering s. farid@ucl. ac. uk ECI Integrated Continuous Biomanufacturing, Barcelona, Spain, 20 -24 October 2013

Acknowledgements Engineering Doctorate Project: Evaluating The Potential of Continuous Processes for Monoclonal Antibodies: Economic, Environmental and Operational Feasibility UCL-Pfizer Collaboration (2008 -2013) James Pollock UCL Suzanne Farid UCL Sa Ho Pfizer UCL academic collaborators included: Daniel Bracewell (ex-)Pfizer collaborators included: Glen Bolton, Jon Coffman Funding: UK EPSRC, Pfizer 2

Bioprocess Decisional Tools – Domain Biotech Drug Development Cycle Farid, 2012, In Biopharmaceutical Production Technology, pp 717 -74 3

Scope of UCL Decisional Tools Typical questions addressed: Process synthesis & facility design Which manufacturing strategy is the most cost-effective? How do the rankings of manufacturing strategies change with scale? Or from clinical to commercial production? Key economic drivers? Economies of scale? Probability of failing to meet cost/demand targets? Robustness? Portfolio management & capacity planning Portfolio selection - Which candidate therapies to select? Capacity sourcing - In-house v CMO production? Impact of company size and phase transition probabilities on choices? 4

Scope of UCL Decisional Tools Systems approach to valuing biotech / cell therapy investment opportunities Ø Process synthesis and facility design Ø Capacity planning Ø Portfolio management Challenges: Capturing process robustness under uncertainty & reconciling conflicting outputs Ø Ø Ø Adopting efficient methods to search large decision spaces Ø Ø Ø Portfolio management & capacity planning (Rajapakse et al, 2006; George & Farid, 2008 a, b) Multi-site long term production planning (Lakhdar et al, 2007; Siganporia et al, 2012) Chromatography sequence and sizing optimisation in multiproduct facilities (Simaria et al, 2012; Allmendinger et al, 2012) Integrating stochastic simulation with advanced multivariate analysis Ø Fed-batch versus perfusion systems (Lim et al, 2005 & 2006; Pollock et al, 2013 a) Continuous chromatography (Pollock et al, 2013 b) Integrated continuous processing (Pollock et al, submitted) Stainless steel versus single-use facilities (Farid et al, 2001, 2005 a &b) Facility limits at high titres (Stonier et al, 2009, 2012) Single-use components for allogeneic cell therapies (Simaria et al, 2013) Prediction of suboptimal facility fit upon tech transfer (Stonier et al, 2013; Yang et al, 2013) Creating suitable data visualization methods Ø For each of above examples Farid, 2012, In Biopharmaceutical Production Technology, pp 717 -74 5

Scope of UCL Decisional Tools Systems approach to valuing biotech / cell therapy investment opportunities Ø Process synthesis and facility design Ø Capacity planning Ø Portfolio management Challenges: Capturing process robustness under uncertainty & reconciling conflicting outputs Ø Fed-batch versus perfusion systems (Pollock et al, 2013 a) Ø Continuous chromatography (Pollock et al, 2013 b) Ø Integrated continuous processing (Pollock et al, submitted) Ø Ø Ø Adopting efficient methods to search large decision spaces Ø Ø Ø Portfolio management & capacity planning (Rajapakse et al, 2006; George & Farid, 2008 a, b) Multi-site long term production planning (Lakhdar et al, 2007; Siganporia et al, 2012) Chromatography sequence and sizing optimisation in multiproduct facilities (Simaria et al, 2012) Integrating stochastic simulation with advanced multivariate analysis Ø Stainless steel versus single-use facilities (Farid et al, 2001, 2005 a &b) Facility limits at high titres (Stonier et al, 2009, 2012) Single-use components for allogeneic cell therapies (Simaria et al, submitted) Prediction of suboptimal facility fit upon tech transfer (Stonier et al, 2013; Yang et al, 2013) Creating suitable data visualization methods Ø For each of above examples Farid, 2012, In Biopharmaceutical Production Technology, pp 717 -74 6

Scope of UCL Decisional Tools Systems approach to valuing biotech / cell therapy investment opportunities Ø Process synthesis and facility design Ø Capacity planning Ø Portfolio management Challenges: Capturing process robustness under uncertainty & reconciling conflicting outputs Ø Fed-batch versus perfusion systems (Pollock et al, 2013 a) § Scenario: New build for commercial m. Ab prodn § Impact of scale on cost § Impact of titre variability and failures rates on robustness Ø Continuous chromatography (Pollock et al, 2013 b) § Scenario: Retrofit for clinical / commercial m. Ab prodn § Impact of scale and development phase on cost § Retrofit costs across development phases Ø Integrated continuous processing (Pollock et al, submitted) § Scenario: New build for clinical / commercial m. Ab prodn § Impact of hybrid batch/continuous USP and DSP combinations § Impact of development phase, company size and portfolio size 7

Fed-batch versus perfusion culture (New build) Ø Fed-batch versus perfusion systems (Pollock et al, 2013 a) § § § Scenario: New build for commercial m. Ab prodn Impact of scale on cost Impact of titre variability and failures rates on robustness Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 8

Fed-batch versus perfusion culture (New build) Commercial products using perfusion cell culture technologies Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 9

Fed-batch versus perfusion culture (New build) Scenario trade-offs: FB v SPIN v ATF Perfusion Spin-filter Perfusion LIQUID LEVEL SPIN FILTER PRO: Investment DSP consumable cost Steady state cell densities Failure rates CON: Equipment failure rate USP consumable cost Scale limitations Validation burden Ø Compare the cost-effectiveness and robustness of fed-batch and perfusion cell culture strategies across a range of titres and production scales for new build Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 10

Fed-batch versus perfusion culture (New build) Key assumptions SPIN ATF Seed #1 Reactor type Seed #2 Cell Culture Suite FB Seed #1 Suites Seed #2 CC CC CC Cent DF FB SPIN ATF SS/SUB SS SUB Cell culture time (days) 12 60 60 Max VCD (106 cells/ml) 10 15 50 Max bioreactor vol. (L) 20, 000 2000 1500 Max perf. rate (vv/day) DF Variable – 1 1. 5 65% 68% 69% 22 5 5 2 – 10 20% FB 45% FB 170 -850 2 x FB 6. 5 x FB Process yield UF Annual # batches Product conc. (g/L) Pro. A VI VI VI Pool CEX UFDF Viral Secure Suite Pro. A CEX DSP Suite Pro. A UFDF VRF VRF AEX AEX UFDF Productivity (mg/L/day) Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 11

Fed-batch versus perfusion culture (New build) Results: Impact of scale on COG = Indirect = Material = Labour Comparison of the cost of goods per gram for an equivalent fed-batch titre of 5 g/L Critical cell density difference for ATF to compete with FB - x 3 fold. Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 12

Fed-batch versus perfusion culture (New build) Uncertainties and failure rates Process event p(Failure) Consequence Fed-batch culture contamination 1% Spin-filter culture contamination 6% Spin-filter failure 4% ATF culture contamination 6% ATF filter failure 2% In process filtration failure – general 5% 4 hour delay & 2% yield loss 20 % 4 hour delay & 2% yield loss In process filtration failure– post viral inactivation Batch loss & discard two pooled perfusate volumes Batch loss & no pooled volumes are discarded Batch loss & discard two pooled perfusate volumes Replace filter & discard next 24 hours of perfusate Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 13

Fed-batch versus perfusion culture (New build) Results: Impact of variability on robustness Annual throughput and COG distributions under uncertainty 500 kg demand, 5 g/L titre Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 14

Fed-batch versus perfusion culture (New build) Results: Impact of variability on robustness Annual throughput and COG distributions under uncertainty 500 kg demand, 5 g/L titre Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 15

Fed-batch versus perfusion culture (New build) Results: Reconciling operational and economic benefits Operational benefits dominate 1. FB 2. ATF 3. SPIN 1. FB = ATF 2. SPIN 1. ATF 2. FB 3. SPIN Economic benefits dominate ─ fed-batch, -- spin-filter, ··· ATF Pollock, Ho & Farid, 2013, Biotech Bioeng, 110(1): 206– 219 16

Continuous chrom: clinical & commercial (Retrofit) Ø Continuous chromatography (Pollock et al, 2013 b) § Scenario: Retrofit for clinical / commercial m. Ab prodn § Impact of scale and development phase on cost § Retrofit costs across development phases Pollock, Bolton, Coffman, Ho, Bracewell, Farid, 2013, J Chrom A, 1284: 17 -27 17

Continuous chrom: clinical & commercial (Retrofit) Technology Evaluation 1 ml scale-down evaluation 3 C-PCC system validation Load Discrete event simulation tool Wash Load m. Ab Breakthrough 100% 80% 60% 40% 100 cm/hr (14. 3 mins) 230 cm/hr (6. 6 mins) 20% 300 cm/hr (5 mins) 500 cm/hr (3 mins) 0% 0 FT 50 100 Challenge Load (mg/ml) FT 18 150 Mass balance, scale-up & scheduling equations 18

Continuous chrom: clinical & commercial (Retrofit) Example Chromatogram ramp-up Switch time ramp-down 3 C-PCC CV = 3 x 1 m. L Titre = 2 g/L tres = 6. 6 mins t. Switch = 200 mins trampup = 330 mins trampdown = 300 mins 19 19

Continuous chrom: clinical & commercial (Retrofit) Product Quality (Elution peak) CEX - HPLC Cycle (100 cycles) Batch (3 cycles) 3 C-PCC (6 runs) Acidic 19. 3 % 18. 4 % 18. 3 % Designated 75. 0 % 74. 7 % 75. 8 % Basic 5. 7 % 6. 8 % 5. 9 % HMW 0. 7 % 1. 0 % 0. 4 % Designated 97. 6 % 96. 9 % 98. 0 % LMW 1. 7 % 2. 1 % 1. 6 % SEC - HPLC Cycle (100 cycles) Batch (3 Cycles) 3 C-PCC (6 runs) 20 20

Continuous chrom: clinical & commercial (Retrofit) Technology Evaluation 1 ml scale-down evaluation 3 C-PCC system validation Load Discrete event simulation tool Wash Load m. Ab Breakthrough 100% 80% 60% 40% 100 cm/hr (14. 3 mins) 230 cm/hr (6. 6 mins) 20% 300 cm/hr (5 mins) 500 cm/hr (3 mins) 0% 0 FT 50 100 Challenge Load (mg/ml) FT 21 150 Mass balance, scale-up & scheduling equations 21

Continuous chrom: clinical & commercial (Retrofit) Early phase DS manufacture challenges Proof-of-concept (Phase I & II) ~ 4 kg DS for the average m. Ab 1, 2 1800 L (wv) Fed-batch @ 2. 5 g/L Day 1 PA (1 cycle) Day 2 Day 3 PA (2 cycle) Day 4 AEX Day 5 VRF Day 6 UFDF Protein A resin costs ~ 60% Direct manufacturing costs ~ $250 k per molecule 1. Simaria, Turner & Farid, 2012, Biochem Eng J, 69, 144 -154 2. Bernstein, D. F. ; Hamrell, M. R. Drug Inf. J. 2000, 34, 909– 917. Pollock, Bolton, Coffman, Ho, Bracewell, Farid, 2013, J Chrom A, 1284: 17 -27 22

Continuous chrom: clinical & commercial (Retrofit) Results: Economic Impact – Protein A Proof-of-concept (Phase I & II) ~ 4 kg DS for the average m. Ab (2. 5 g/L) Standard 3 C-PCC Load Wash 31. 4 L 3 x 4. 9 L = 14. 7 L 5 cycles 17 cycles $ 250 K resin $ 118 K resin 8 hour shift 24 hour shift 53% reduction in resin volume 40% reduction in buffer volume x 2. 3 increase in man-hours 23

Continuous chrom: clinical & commercial (Retrofit) Results: Impact of scale on direct costs PA costs Other Costs 1 x 4 kg 4 x 10 kg 20 x 10 kg Pollock, Bolton, Coffman, Ho, Bracewell, Farid, 2013, J Chrom A, 1284: 17 -27 24

Continuous chrom: clinical & commercial (Retrofit) Results: Impact of development phase on retrofitting investment Po. C (1 x 4 kg) PIII & Commercial (4 x 10 kg) STD: ÄKTA process (15 -600 L/hr) + 0. 4 m column x 4 Investment 4 C-PCC (15 -600 L/hr) + 4 x 0. 2 m columns ~8 Po. C batches STD: ÄKTA process (451800 L/hr) + 0. 5 m column 4 C-PCC (15 -600 L/hr) + 4 x 0. 3 m columns x 3. 3 Investment ~25 PIII batches or ~ 8 Po. C batches 25

Integrated continuous processes (New build) Scenarios: Alternative integrated USP and DSP flowsheets Ø Integrated continuous processing (Pollock et al, submitted) § § § Scenario: New build for clinical / commercial m. Ab prodn Impact of hybrid batch/continuous USP and DSP combinations Impact of development phase, company size and portfolio size DSP scheduling a) batch process sequence b) continuous + batch process sequence c) continuous process sequence Pollock, Ho & Farid, submitted 26

Integrated continuous processes (New build) Results: Impact of development phase and company size on optimal Strategies USP Capture Polishing Base case FB-CB ATF-CB FB-CC ATF-CC Fed-batch ATF perfusion Batch Continuous Batch Continuous USP + Continuous Capture + Continuous Polishing Batch USP + Continuous Capture + Batch Polishing 27

Summary Process economics case study insights: • Fed-batch versus perfusion culture for new build – Economic competitiveness of perfusion depends on cell density increase achievable and failure rate • Continuous chromatography retrofit – Continuous capture can offer more significant savings in early-stage clinical manufacture than late-stage • Integrated continuous processes for new build – Integrated continuous processes offer savings for smaller portfolio sizes and early phase processes – Hybrid processes (Batch USP, Continuous Chrom) can be more economical for larger / late phase portfolios 28

UCL Decisional Tools Research Operational & Economic Evaluation of Integrated Continuous Biomanufacturing Strategies for Clinical & Commercial m. Ab Production Suzanne Farid Ph. D CEng FIChem. E Reader (Associate Professor) Co-Director EPSRC Centre for Innovative Manufacturing UCL Biochemical Engineering s. farid@ucl. ac. uk ECI Integrated Continuous Biomanufacturing, Barcelona, Spain, 20 -24 October 2013

Backup 31

Continuous chrom: clinical & commercial (Retrofit) 3 Column Periodic Counter Current Chromatography Load FT Wash/ Elution FT Pollock, Bolton, Coffman, Ho, Bracewell, Farid, 2013, J Chrom A, 1284: 17 -27 32

Continuous chrom: clinical & commercial (Retrofit) 3 Column Periodic Counter Current Chromatography 40 g/L Wash/ Elution Load FT 65 g/L Wash FT FT Pollock, Bolton, Coffman, Ho, Bracewell, Farid, 2013, J Chrom A, 1284: 17 -27 33

Continuous chrom: clinical & commercial (Retrofit) Results: Environmental Impact Proof-of-concept (Phase I & II) ~ 4 kg DS for the average m. Ab (2. 5 g/L) STD 3 C-PCC -40% e-factor (kg/ kg of protein) STD 3 C-PCC Difference Water 5900 5250 -11% Consumable 24. 5 13. 7 -44% Pollock, Bolton, Coffman, Ho, Bracewell, Farid, 2013, J Chrom A, 1284: 17 -27 34

Integrated continuous processes (New build) Results: Impact of development phase and company size on optimal Strategies USP Capture Base case FB-CB ATF-CB FB-CC ATF-CC Fed-batch ATF perfusion Batch Continuous Batch USP + Continuous Capture Continuous USP + Continuous Capture 35

Impact of Resin Life Span (Mab. Select x 100 cycles) • Standard cycling study (40 mg/ml) 19% loss in capacity • Column regeneration (Na. OH) 12% loss in capacity • 100% breakthrough cycling study – x 2. 2 the load volume vs. standard 30% loss in capacity Insignificant loss < 15 cycles 36 36

Commercial Manufacture Feasibility (3 C-PCC @ 5 g/L) Increasing cycle number 16 22 Increasing cycle number 38 16 Batch 11 – surpasses harvest hold time 19 38 Batch 6 – surpasses pool vessel volume 37 37