Title: Development and Applications of PAT in API Unit Operations
1Development and Applications of PAT in API Unit
Operations
San Kiang September 11, 2006 Bristol-Myers
Squibb New Brunswick, NJ
2Overview
- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- Case Studies
- Summary
3Pharmaceutical Industry Commercial Challenges
- Current cost of bringing new drug to market
gt800 million - Success rate 1 in 10
- Time to Market 10 to 15 years
- Highly competitive
- Highly regulated
4Regulatory Challenges ICH Q8 Guidance
- Demonstrate a higher degree of understanding of
manufacturing processes and process control for
more flexible regulatory approaches - Risked based regulatory decisions (reviews
inspections) - Process improvements within the design space
without further regulatory review - Real time quality control to reduce end-product
release testing
Courtesy FDA/CDER
5Regulatory ChallengesNew Pharmaceutical Quality
Assessment System (PQAS)
Objective To Ensure that Necessary Quality
Attributes are Built in (QbD) and the Drug
Product can be Manufactured Consistently with
High Quality for its Intended Use (i.e., Safety
and Efficacy)
- Submissions rich in scientific knowledge
demonstrating understanding of product and
process - Specifications set based on product requirements
for safety, efficacy, and stability - Process designed and controlled to robustly and
reproducibly deliver quality product
Moheb Nasr, FDA
6- Industry Challenges
- New API Development Paradigm
- Quality by Design
- Process Work Flow
- Strategic Benefits in API Unit Operations
- Enabling Technology Tools
- Case Studies
- Summary
7Work Smarter, Not HarderStrategy for a New
Paradigm
Meet the Challenges by executing fundamental
changes to the development and manufacturing
approach
Integrated scientific, manufacturing,
commercial objectives Development of the best
route with full intrinsic process
knowledge Apply new technologies for effective
utilization of process knowledge
Integration
Science
Manufacturing
Bottom Line How to Achieve these ?
8Fundamental Approach in Development Paradigm
Shift
Quality by Design
Traditional Approach
Real Time Control for Continuous Improvement
Causal Links Predict Performance
Scale Up Prediction
Modeling for Mechanistic Understanding
Decision Based On Univariate Approach
PAT
DOE (Multivariate Systems Approach)
Data Derived From Trial Error Experimentation
Identification of PCCP
PCCP Process Critical Control Parameter
9Quality by Design (QbD)21st Century cGMP
Manufacturing
- The product is designed to meet intended use
- The process is designed to consistently meet
product critical quality attributes - The impact of starting materials and process
parameters on product quality is understood - Critical sources of process variability (raw
materials, process) are identified and controlled - The process is continually monitored and updated
to allow for consistent quality over time
Moheb Nasr, FDA
10Road Map to QbD
Designing for the API Manufacturing
Quality By Design
Chemistry Selection
Process Modeling
Process Devel.
Process Control
Process Scale Up
Manufacturing Science
Science of Design
Evaluate
Design
Predict
Verify
Control
Deliverables
Best Production Chemistry
Intrinsic Process Knowledge
Optimized Unit Operations
Demonstrated Production Process
Real Time Continuous Improvements
11API Process Development Workflow in 21st Century
Molecular model of physical properties, kinetics,
selectivity, solvent effects, and transport
data Disciplined route selection
Form selection
Form screening
Molecular and unit operation models guide
experimental design. Determining design space
early on
Route feasibility
Route selection
Unit operation model on reaction, distillation,
crystallization, and drying Disciplined process
selection
Process dev. studies
Process selection
Develop calibrate Model
Plant data to verify model
Integrated reactor studies
Model qualification 1
PAT Thermo-dynamics data
Glass plant
Pilot plant
Model qualification 2
Tech transfer Manufac. imp.
Predict scale up with qualified model Model
design space QbD
API
Equipment Train capability model, Equipment
capability, (Unit Operation Model)
12- Industry Challenges
- New API Development Paradigm
- Quality by Design
- Process Work Flow
- Strategic Benefits in API Unit Operations
- Enabling Technology Tools
- Case Studies
- Summary
13API Manufacturing - Typical Unit Operations
PAT
Raw Material
Reaction
Purification
Crystallization
Isolation Drying
API
Right API Properties Enables Formulation
of Quality Drug Product
- Drug Product
- Drug-excipient compatibility
- Stability
- Dosage form design
- Bioavailability
- Manufacturability
14Impact of API Powder Properties on Pharmaceutical
Product
Inhomogeneous Distribution of API with Excipient
Wide PSD
Poor Dissolution Performance
RAMAN m
SEM
size
Homogeneous Distribution of API with Excipient
Narrow PSD
Good Dissolution Performance
RAMAN m
size
SEM
Excipient
API
Narrower particle size distributions (PSD)
minimize segregation problems during
mixing, rendering a more homogeneous distribution
of components in the final product
15Impact of On-line Process Analyzers - Reaction
PAT
Raw Material
Reaction
Purification
Crystallization
Isolation Drying
API
- Safety
- Minimize handling
- Tight hazardous process control
- Quality/Productivity
- Mechanistic kinetic knowledge
- Parametric boundary
- Access to extreme conditions
- Real time monitoring control
- Continuous quality assurance
16Impact of On-line Process Analyzers - Downstream
Process
PAT
Raw Material
Reaction
Purification
Crystallization
Isolation Drying
API
- Quality
- Process knowledge
- Parametric boundary
- Real time monitoring control
- Continuous quality assurance
- Productivity
- Seamless scale up
- Minimize OOS product
- Rapid troubleshoot
- Optimal process efficiency
- Safety
- Minimize handling
- exposure
17- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- PAT
- Predictive Modeling
- Risk Analysis
- Integrated Technology Solutions
- Case Studies
- Summary
18PAT Tools Driving Enabling Quality by Design
Process Knowledge
Critical Parameters Identified
Explained Variability Managed by Process Product
Quality Attributes Reliability Predicted Continuou
s Quality Assurance
Multivariate Data Acquisition Analysis
On-Line Sensors for Monitoring Control
Process Modeling
Risk Analysis
19Applications of On-Line Process Analyzers during
Crystallization
FBRM Probe
- FBRM to monitor control Particle Size
Distribution - Isolation/Drying Efficiency
- Powder properties
- Formulation performance
Frequency (number of particles)
size (microns)
20PAT Tools Enabling Quality by Design
Process Knowledge
Critical Parameters Identified
Explained Variability Managed by Process Product
Quality Attributes Reliability Predicted Continuou
s Quality Assurance
Multivariate Data Acquisition Analysis
On-Line Sensors for Monitoring Control
Process Modeling
Risk Analysis
21Science-Risk Based Approach - Learn Predict for
Optimized Process
Predictive Modeling Risk Analysis
Experimental Studies
Theoretical Analysis
- Direct Experiment
- Correlate Experimental Results
- Predict Scale Up Performance
- Determine Parametric Sensitivities
- Risk Management
- Identify Critical Process Parameters
- Design Scale Down System
- Collect Process Data
22Applications of Modeling in BMS
Molecular Modeling
System Modeling
Unit Operation Modeling
- Molecular Structure
- Relative Reactivity Selectivity
- Reaction Pathway Mechanism
- Thermochemistry (Heat of Reaction)
- Relative pKa Value of Intermediates
- Physical Properties
- Homogeneous Heterogeneous Catalyst Selection
- Polymorph Selection
- Equilibrium Kinetic Knowledge
- Prediction of Process Results
- Prediction of Scale Up Performance
- Chemical Route Selection
- Process Synthesis
- Process Options Selection
- Cost Analysis for Metrics
- Optimal Scheduling for PRD Pilot Facilities
- Optimization of Equipment Use for Multi-batch
Campaigns - Plant Fit Determination Selection of
Manufacturing Site
Mixing Reaction Crystallization Extraction Chromat
ography Distillation
23- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- PAT
- Predictive Modeling
- Risk Analysis
- Integrated Technology Solutions
- Case Studies
- Summary
24PAT Tools Enabling Quality by Design
Process Knowledge
Critical Parameters Identified
Explained Variability Managed by Process Product
Quality Attributes Reliability Predicted Continuou
s Quality Assurance
Multivariate Data Acquisition Analysis
On-Line Sensors for Monitoring Control
Process Modeling
Risk Analysis
25Risk Management ConceptICH Q9 Quality Risk
Management
Risk Assessment
Risk Identification
Risk Analysis
Risk Evaluation
unacceptable
Risk Control
Risk Reduction
Risk Acceptance
Risk Communication
Risk Review
acceptable
Risk Communication
Risk Acceptance
Output of the Risk Management Process
Review Events
26Risk Assessment Tools
27FMEA Analysis Applied to Pharmaceutical Process
Critical parameter identification analysis
FMEA execution based on Critical Parameters
Evaluation of risk reduction methods
All RPN gt RPNacceptable
Risk Prioritization Number (RPN) ranking
All RPN lt RPNacceptable
Communication of FMEA results
28Severity, Probability and Detection evaluated for
each of the subsystems in System 1 (Final step to
prepare API)
29Implementation of risk reduction steps and FMEA
for RPNs above acceptable RPN
Final RPNs
30- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- PAT
- Predictive Modeling
- Risk Analysis
- Integrated Technology Solutions
- Case Studies
- Summary
31Technology Solution - Road Map to Design a
Better Process
Particle Engineering
PAT
- Collect Fundamental Data
- Physicochemical properties
- Kinetics Equilibrium
- Mixing/Diffusion Effect
- Mass Transfer Effect
- Heat Transfer Effect
Linked
Statistical Design
Automation
Intrinsic Process Knowledge
- Formulation
- Critical formulation parameters
- Optimal formulation
- Physicochemical properties
- Delivery technologies
- Process
- Critical process parameters
- Mechanistic Kinetics
- Physicochemical properties
- Manufacturability
32Material Science - API Influence on Drug Product
Integration of On-line Process Analyzers for
Consistent Physicochemical Properties
Crystal SizeCrystal Shape Polymorph
Control Cycle Time Process Design Equipment
Selection
Cycle TimeEquipment Selection
Flowability Compactibility Bioavailability
Formulation Dosage Manufacturability
API
Particle Size Distribution
Supersaturation Kinetics
Drying
Filtration
Crystallization
33API Properties Controlled by Different Unit
Operations
- Crystallization
- Form
- Particle Size Distribution
- Morphology
- Filtration/Drying
- Attrition/Agglomeration
- Form
- Milling
- Particle Size Distribution
- (de-glomeration, size reduction)
34Particle Engineering - API Influence on Drug
Product
Agglomerated Particles
Attrited Particles
35Particle Engineering API Influence on Tablet
Performance
Original procedure Large primary particles
150-300 um
Poor toughness chipped tablets
100 ?m
Modified procedure (High S) Agglomerates, good
flowability
Optimized procedure 10-20um primary particles
excellent toughness
500 ?m
36- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- Case Studies
- Reaction
- Hydrogenation
- Crystallization
- Summary
37Case Study Nucleophilic Addition
Stoichiometry
SM1 1.1 SM2 3.2 C 2.5 D ? Prod 0.1 F
3.3 G 1.1 H 1.5 I
Impurities (non-stoichiometric)
?
Dimer SM2Dimer IMP1
Major reactions
38Inline Monitoring
- Real-time concentrations of species,
- including some reaction intermediates
- Large quantity of precise kinetic data
- Calibrated by HPLC of stable species
Raman
FTIR
39Byproduct Formation
PAT and model show good agreement
Coupling Agent Degradation (Slow)
k5 0.0033 L3/2/mol3/2s
SM2 C ? J H ? Imp1 G J SM2 ?
SM2Dimer H
C
k5
SM2
J
r a Sm22C1/2 (high base conc.) r a Sm2
C (low base conc.)
k6 4.8 x 10-5 L/mols
Series Coupling to Form Dimer (Slower)
Product
k6
Prod SM2 ? Dimer H
SM2
Dimer
r k6ProdSM2
40Complete Model Synthesis
- The main reaction steps have much higher rate
constants than the byproduct steps - The deprotonation steps have rate constants
consistent with the acidity of the proton removed
(i.e. k1 gt k4 gt k2)
- The Rate Limiting Step is either the second
depro-tonation of SM2 - (Step 2) or the coupling
reaction (Step 3)
SM2
Prod
Dimer
Full model prediction is in agreement w. exp
G
SM1
41Case Study Summary
- Kinetics experiments used online FTIR
spectroscopy for mechanistic investigation and
quantitative data collection - BatchCAD kinetic model has been developed to
provide a more complete understanding of the
overall reaction and byproduct pathways - Six-step kinetic mechanism
- formation of desired products through
deprotonation and coupling - degradation of SM2
- coupling with Product to produce Dimer
- Model provides a way to simulate operating ranges
and inputs for FMEA for risk assessment of common
operating issues - fast base addition
- under and over-charging of reactants
42- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- Case Studies
- Reaction
- Hydrogenation
- Crystallization
- Summary
43Case Study Carbobenzyloxy Deprotection Reaction
- Spec. for Reaction end point 0.6 RAP of RNHCBz
- Removal of CO2
THF
FTIR
Hydrogenation
THF 10 Catalyst Loading
- End Point Determination - rate of impurity
formation increases 25 after the desired
reaction is complete - Hydrogen uptake measurement is not reliable to
monitor the kinetics of the reaction since CO2 is
evolved as a by-product - CO2 removal upon reaction completion to ensure
product stability
44Kinetic Understanding of CBz Deprotection - CO2
Effects
CO2 poisons the catalyst. Reaction rate depends
on changes of the reactor headspace due to CO2
amount variations in liquid
Reaction changing from first to zero order as
headspace is changed Reaction is zero order in
substrate.
45Chemometric Model for Reaction End point and CO2
Removal
CO2 in the liquid as a function of headspace in
the absence of catalyst at 25 oC.
(1746cm-1 1696.62cm-1)
- Considerations
- Kinetics of the reaction
- VLE of CO2
46Validation of Model
- Model validated on scale up data
- RSD value is 1.43
- Good agreement between experimental data and
quant model predictions
47- Industry Challenges
- New API Development Paradigm
- Enabling Technology Tools
- Case Studies
- Reaction
- Hydrogenation
- Crystallization
- Summary
48Case Study Improving Bioavailability of BMS-XXX
Issue Aqueous solubility of only 0.04 mg/mL at
pH 7
In process solvents
Polymorphs of BMS-XXX
Solvates or Hydrates Dihydrate H2-2 AN-3 DS-3 0.5P
G-4
Neat Form N-1
Desired Form
- Batch Crystallization to Produce N-1
- Crystal Rod shape with 50 ?m diameter 200 ?m
long - Polymorph Transformation Need 24 hours
- Size Reduction 2 passes in pin-mill for
D9550 ?m
49Case Study Improving Bioavailability of BMS-XXX
N-1 crystals from batch crystallization, unmilled
- Objective
- D95lt 50 ?m to improve bioavailability
- Eliminate multiple milling
- Rugged process for scale up
- Faster polymorph transformation
In process solvents
D95180 ?m
Polymorph Transformation in the process solvent
of PG/H2O
Anhydrous form N-1 (short rods)
Dihydrate H2-2(needles)
D9535 ?m
Transformation from H2-2 to N-1 produce small
crystals
50Particle Engineering Solution -Continuous
Polymorph Transformation
Homogenizer
Small N-1 Crystal
FA
T
H2-2 slurry, RT
T 55-65 ?C ? V/FA 20-25 mins
V
T
- Batch
- D95 200 ?m
- 24 hours
- Multiple Pass Milling
- Continuous
- D95 30 to 40 ?m
- 20 to 25 min
- No Milling
51Model Prediction Validated by Scale Up Data
N1
H2-2
Time( arb)
H2-2
N1
Time
52Summary Seamless Path From Quality Product to
Compliance
Generate Manage Process Knowledge
Quality Drug Product Process
- Strategy
- PAT to generate fundamental data for mechanistic
kinetics understanding - Particle Engineering for designer particles
- Predictive modeling risk analysis
- Real-time control by PAT for continuous process
improvement - Integrated API Drug Product development efforts
- Benefits
- Eliminate potential formulation problems
- Streamline pharmaceutical process operation
- Better product quality control
- Eliminate technical uncertainties for seamless
scale transition - Facilitate Regulatory Filing
53Acknowledgement
- Mauricio Futran
- Yeung Chan
- Jale Muslehiddinoglu
- John Shabaker
- Steve Wang
- Olav Lyngberg
- Soojin Kim
- Chiajen Lai
- Bing-Shiou Yang