Simulated Moving Bed Chromatography in the Pharmaceutical Industry

Simulated Moving Bed Chromatography in the Pharmaceutical Industry Ron Bates Bristol-Myers Squibb April 19, 2004

Outline • Short Biography • What is Bristol-Myers Squibb • Chromatography – Batch vs continuous • HPLC, SMB, P-CAC • Simulated Moving Bed Chromatography – – Introduction Theory (brief) Operation Applications in the Pharmaceutical Industry

• B. S. Chemical Engineering, RPI, 1993 • Ph. D. Biochemical Engineering, University of Maryland, Baltimore County, 1999 – Focus: ion-exchange chromatography • Pfizer, Groton, CT, 1999 -2003 – Focus: small molecule chromatography, HPLC, SFC, SMB, FLASH, extraction, crystallization, precipitation • Bristol-Myers Squibb, Syracuse, NY, 2003 -present – Focus: protein chromatography

Bristol-Myers Squibb • • Top-ten pharmaceutical company Products in numerous therapeutic areas Cardiovascular & Metabolic Diseases Mental Health Pravachol, Coumadin Abilify Headache and Migrane Infectious Diseases Excedrin Reyataz, Sustiva Oncology Erbitux, Taxol • Strong pipeline focused in 10 therapeutic areas – Oncology, Cardiovascular, Infectious Diseases, Inflammation, etc. • Sites around the world – U. S. Research/Manufacturing sites • MA, NY, NJ, CT, IL, Puerto Rico

Bristol-Myers Squibb Syracuse, NY • Clinical and Commercial Manufacturing Plant – Small-molecule pilot plants • Process development and optimization • Clinical manufacturing – Penicillin-based products • Last US-based Penicillin manufacturer – Bio-synthetic products – Biotechnology • Development, Manufacturing, Analytical Biosciences, Quality Control / Assurance

Bristol-Myers Squibb Syracuse, NY - Biotechnology • Two lead protein therapeutics – Abatacept: commericial in 2005 • Commercial-scale manufacturing • Commercial launch out of Syracuse Facility • BLA filing – Dec. 2004 – LEA 29 Y: Phase III clinical trials in 2005 • Development for next generation process • Clinical production in 2004 • Expansion in analytical and quality groups to support processes

Batch vs. Continuous Chromatography

Batch Chromatography • Discrete starting and ending points – Example: 10 minute HPLC cycle – Types: GC, HPLC, FLASH, FPLC, etc. – Can be run in many modes: • Linear, overloaded, frontal, etc.

Batch Chromatography Effluent to Waste Feed Effluent to Waste Load Desorbent Elution (Raffinate) (To Waste) (Extract) Strong Solvent Reference: Linda Wang, Perdue University Regeneration

Batch Chromatography Empty zone

Continuous Chromatography • Feed is loaded onto column and product is collected continuously Feed column • Annular (P-CAC) – Preparative continuous annular chromatography • Countercurrent – Simulated moving bed chromatography (SMB)

P-CAC Reference: Genetic Engineering News, Oct. 1, 1999

Simulated Moving Bed Chromatography (SMB)

What is SMB • SMB is Simulated Moving Bed Chromatography. • SMB is continuous countercurrent chromatography. The feed is pumped into the system and two (or more) product streams are continuously collected. • SMB has been used for the production of millions of tons of bulk commodities (p-xylene, high fructose corn syrup, etc. . . ) for the past four decades. • Due to improvements in column and equipment technology, SMB has recently been used in the pharmaceutical industry (Sandoz, Smith. Kline, UCB, Pfizer). – – HPLC costs: $100/kg to $5000/kg SMB costs: $50/kg to $200/kg

SMB versus HPLC Advantages of SMB: – Lower solvent utilization (up to 10 times less than batch HPLC) – Generally can use less expensive, larger stationary phases – Able to get high recovery and high purity – Sometimes better productivity – Lower labor and QC costs – Only partial separation of solutes is required to obtain high purity. – Higher yield than batch – 10% more than batch. – High throughput – 5 to 10 fold increase. – Lower solvent consumption – An order of magnitude lower. – Continuous process. Disadvantage of SMB: – Binary separation only – Complexity

Commercial Applications of SMB • • • Hydrocarbons Sugars Agrochemicals Antibiotics Peptides Chiral Drugs – Gaining tremendous momentum – FDA approves of the technology – Chiral resin manufacturers sell resins specifically made for SMB • Proteins? – Useful as polishing step? • SEC: remove aggregated form of product – Multicomponent separations more difficult than traditional uses • 8, 12, even 16 zone systems being examined

Continuous Countercurrent Chromatography Basic Principle Feed stationary column Mobile Phase A sample is injected in the centre of a stationary column The two components move at different speeds and are separated If we now move the column from right to left, at a speed halfway between that of the solutes, they now move in different directions. . .

Continuous Countercurrent Chromatography Basic Principle column Feed Mobile Phase The two solutes now move in different directions relative to a stationary observer. If the column is very long, the bands will continue to separate.

Continuous Countercurrent Chromatography Basic Principle column Feed Mobile Phase If we continue to add sample at the center, the components will continue to separate

Continuous Countercurrent Chromatography Basic Principle column Feed Mobile Phase This is clearly a continuous system, but there are problems. The column needs to be of infinite length, the actual moving of solids is very difficult and some way to introduce and remove the sample and the products are needed. We solve this by cutting the column into small segments and simulating the moving of them

Continuous Countercurrent Chromatography Basic Principle column Feed Mobile Phase The feed and solvent inlets are now placed between the segments and are moved each time a segment is moved from one end to the other

Continuous Countercurrent Chromatography Basic Principle column Mobile Phase Feed Mobile Phase Products are removed by bleeding off a carefully calculated flow at suitable exit points. This changes the velocity of the bands in the column and forces the products to move toward the ports This ensures that the column segments are clean before they are moved and that the solvent can be recycled directly back through the system

True Moving Bed

Binary Separation in a True Moving Bed Raffinate Time : t Desorbent Feed Extract Feed Time : t + t Raffinate Extract Desorbent Reference: Linda Wang, Perdue University

Binary Separation in a True Moving Bed Extract Time : t + 2 t Feed Desorbent Raffinate Desorbent Time : t + 3 t Raffinate Extract Feed Reference: Linda Wang, Perdue University

Binary Separation in a True Moving Bed Raffinate Time : t + 4 t Desorbent Feed Extract Feed Time : t + 5 t Raffinate Extract Desorbent Reference: Linda Wang, Perdue University

TMB to SMB • Since it’s very difficult to move solids, true countercurrent chromatography does not exist. • Instead, the bed is broken into many fractions and their movement is simulated by changing the inlet and outlet ports

Simplified SMB - 1 Feed Solvent 1 Extract Solvent 2 3 4 The system is started. . . Raffinate Feed A frontal elution separation occurs in Section 3. Extract Raffinate

Simplified SMB - 2 Solvent Feed The separation continues. . . Extract Solvent Extract Raffinate Feed Raffinate Eventually the front of pure product 1 reaches the outlet. It is distributed between the final Section and the product port

Simplified SMB - 3 Solvent Feed Extract Solvent Extract Raffinate Feed Raffinate Finally, the mixed product reaches the outlet. To avoid collecting impure material, it is necessary to move the columns 1 position upstream.

Simplified SMB - 4 Solvent Feed The frontal separation continues; at the same time, the slow moving product starts to separate from the tail of the mixed product band in Section 2 Extract Solvent Raffinate Feed Eventually the fast moving product again reaches the outlet and more pure product is collected. Extract Raffinate

Simplified SMB - 5 Solvent Feed Raffinate Extract Solvent When the mixed band reaches the end of Section 3 its tail has left Section 2 (if the separation has been correctly designed) and only pure product 2 remains in Section 2. Feed To avoid collecting impure raffinate, the columns are moved once more. Now, the pure component 2 is in Section 1. Extract Raffinate

Simplified SMB - 6 Solvent Feed The second component is now collected at the Extract port while the separation continues in Sections Raffinate 2 and 3. Extract Solvent Extract Feed The faster component reaches the Raffinate port and is again collected; note that the outlet concentrations are neither constant nor concurrent. Raffinate

Simplified SMB - 7 Solvent Feed Eventually, the mixed zone reaches the raffinate port and the columns are again switched. Raffinate Extract Solvent Switch Feed This simplified system is now in a steady state mode and will continue to cycle. Extract Raffinate

• The moving of the bed is simulated by moving the points of feed and mobile phase addition, as well as the points of raffinate and extract removal while keeping the column positions fixed. Mobile Phase Extract Time = 0 Packed Column Raffinate Feed Mobile Phase Extract Time = 1 Raffinate Feed

SMB Configurations The zones are made up of one or more columns • Six-column SMB System I II III IV • Eight-column SMB system I II III IV

SMB Operation

Theory – Governing Equations For another day… Maybe

Theory – Working Equations / Definitions • k’ 1 = capacity factor = (tr-t 0) / t 0 • α = k’ 2 / k’ 1 • Rs = 2* (tr 1 -tr 2) / (w 1 -w 2)

SMB – Method Development 1. Start with linear batch experiments 2. Increase either mass or volume of load to overload the column 3. Calculate isotherm 4. Determine resistance to mass transfer (if important) 5. Calculate necessary flow rates 6. Optimize (either on-the-fly or with a proven model)

Linear Chromatography

Batch Chromatography Experiments • Feed concentration – As concentrated as possible to minimize disruption to Zone III velocity – Need to run batch experiments at appropriate concentrations and solvents • Desorbent composition – Solubility of products – Strength • Trade-off between time and mobile phase utilization • Sorbent – Capacity, selectivity, resolving power

Feed Concentration Feed concentration: Consider two systems – A: Concentrated feed – B: Dilute feed Run batch experiments to examine effect of concentration

Desorbent composition Multiple trade-offs: • Solubility of products and effectiveness of the solvent – Not always complimentary – Often solubility dictates solvent composition • Speed – Low k’ = high throughput • More wear and tear on equipment • Larger system needed – Large k’ = low throughput • Less wear and tear • Smaller system acceptable

Choice of Sorbent • Capacity: higher = better? • Selectivity: higher α = better? • Resolving power: higher Rs = better?

Linear Chromatography

Volume Overloading

Batch Chromatography to SMB Initial Operating Conditions • Determine optimal feed concentration, stationary phase and mobile phase composition (highest α with lowest capacity factors) • Calculate isotherm and mass transfer resistances • Either use software package or rules of thumb to generate initial SMB flow rates

Solvent Mass Balances – Flow Rates v. Recycle I v. D v. I II v. X Zone velocities • v. I = v. Recycle + v. D • v. II = v. I - v. X • v. III = v. II + v. F • v. Recycle = v. III - v. Raff v. II III v. F v. III IV v. Raff Overall Mass Balance • v. D + v. F = v. X + v. Raff

Flow rates • Commercial SMB design models available – Given batch results from 5 -10 column experiments • Flow rate, feed concentrations, retention times • Solubility data – Predict zone velocities, productivities – Problems: • Usually assumes simple adsorption model and lumped mass transfer coefficients • Often difficult to interpret overloaded chromatograms • Rules of Thumb – Educated guesses based upon batch results from linear and overloaded experiments • VII and VIII ratio (based upon retention times) • VI to flush back-side of slowest component from zone I • Feed concentration and flow rate based upon solubility data and solvent mass balance

Period • The period is the time a column stays in one zone also called switching time. • Changing the period has the effect of changing all 4 zones simultaneously, thus either speeding up or slowing down the solutes

Example of switching time

SMB Optimization • Independent variables: – Flow rates • Recycle, Desorbent, Raffinate, Extract, Feed – Period (switching time) – That’s it. • Procedure: – Get the system bound, manipulate the flow rates to maximize throughput at required purity

SMB Optimization v. Recycle I v. D v. I II v. X v. II III v. F v. III IV v. Raff Questions: • What is the effect of increasing the Zone I flow rate? – How would one accomplish this? • Zone II? Zone III? • What if the system is underutilized (i. e. , more feed can be added to the system) – how would one do this without affecting the other zone flow rates?

Two component SMB System Feed Desorbent II IV III Conc. I Bed Position Extract Raffinate

SMB Optimization v. Recycle I v. D v. I II v. X v. II III v. F v. III IV v. Raff Questions: Extract contains too much of the weakly adsorbed species – what do you do? If situation was reversed?

Two component SMB System Feed Desorbent II IV III Conc. I Bed Position Extract Raffinate

SMB Optimization v. Recycle I v. D v. I II v. X v. II III v. F v. III IV v. Raff Questions: Extract contains too much of the weakly adsorbed species – what do you do? If situation was reversed?

Two component SMB System Feed Desorbent II IV III Conc. I Bed Position Extract Raffinate

Examples of SMB

Two component SMB System

Multi-component System 0. 8 Sulfuric Acid Glucose Xylose Acetic Acid 0. 7 0. 6 Ci/CF, i 0. 5 0. 4 0. 3 0. 2 0. 1 0 0 10 20 Time [min] Single-component pulse data Reference: Linda Wang, Perdue University 30 40

Multi-Component SMB System Feed (1, 2, 3) Concentration Desorbent II I Extract (2, 3) Bed Position 1 Fast Solute 2 Intermediate Solute 3 Slow Solute Reference: Linda Wang, Perdue University III IV Raffinate (1)

Complete Separation in Tandem SMB Des. 1 Ext. Feed Raf. i C /C F, i Sulfuric Acid Glucose Acetic Acid 0. 5 0 0 5 Des. Ext. 15 Feed 20 Raf. i C /C F, i 1 10 0. 5 0 0 Reference: Linda Wang, Perdue University 5 10 Column Number 15 20

Profiles of a Parallel SMB I 1. 2 1 D 1 ¯ II E 1 III IV B(o) F ¯ V VI R 1 D 2 ¯ VIII IX E 2 B(i) R 2 ¯ i C /C F, i 0. 8 0. 6 Þ 0. 4 Ü Þ 0. 2 Þ Þ 0 0 5 * Ü Ü Þ * 10 Ü 15 Column Number Glucose yield: 94% Reference: Linda Wang, Perdue University Glucose purity: 99% 20 Sulfuric Acid Glucose Acetic Acid