Design and Deployment of Risk Management Tools to support QbD and Control Strategy Construction

Transcription

1 Design and Deployment of Risk Management Tools to support QbD and Control Strategy Construction Bert Frohlich, Director, Process Controls and Quality by Design Manufacturing Science & Technology, Shire Pharmaceuticacls, 300 Shire Way, Lexington, MA USA International Forum on Process Analytical Chemistry (IFPAC Annual Meeting, 2016 Our purpose We enable people with life-altering conditions to lead better lives 1

2 Design and Deployment of Risk Management Tools to support QbD and Control Strategy Construction Abstract A comprehensive risk management methodology was designed and is being implemented at Shire to support a new QbD-oriented work flow in process development. The workflow culminates in a risk-based control strategy that accompanies a manufacturing process as it is transferred into a commercial facility operating under GMP. The approach developed attempts to use quantitative and semiquantitative measures of risk where ever possible and which are used throughout the QbD workflow to encourage consistency and efficiency through reuse of terms and standardized worksheets. To accommodate knowledge management (KM) and provide justifications for risk assessments, software tools are under evaluation as a more robust KM system than the spreadsheet-based prototype that has been developed. To be built on a database, this tool will support and archive all process assessments along the development pathway and will continue to support a process in manufacturing throughout its lifecycle. Experiences and challenges to the implementation of this new approach will also be shared in this presentation. Bert Frohlich, Director Process Controls and Quality by Design, Manufacturing Science & Technology, Shire Pharmaceuticals, 300 Shire Way, Lexington, MA USA

3 Design and Deployment of Risk Management Tools to support QbD and Control Strategy Construction Outline Goals and Objectives Work Flow Design Risk Assessment Methods Control Strategy Construction Lessons Learned so far 3

4 Why QbD? Risk Prioritization! The Biopharmaceutical Paradox!! Speed to Market Speed to Market First to market reward Market share Lower investment risks Investor relations Limited resources vs. Manufacturing Process Performance / Robustness Process Robustness Thorough product and process understanding Design space definition Less implementation risk Lower cost and failure rate Lower compliance risk 4

5 Strategic Direction Objectives and Benefits.. Mission: Adopt Quality-by-design principles and methods that are business-appropriate for Shire biopharmaceutical process development, design, technology transfer and implementation. Objectives: Define and establish Shire-appropriate QbD approach Clarify business process for QbD-based development and tech transfer to improve consistency of approach Standardize documentation to promote consistency between PD project teams and functions Provide means of demonstrating and documenting product and process knowledge and understanding Standardize risk assessment methods Address regulatory expectations Embrace risk management practices Extent of QbD filing? (out of scope currently future work) 5

6 QbD Work Flow Design Major Execution Stages in and Product / Process Lifecycle PRODUCT DEVELOPMENT QbD principles extend beyond development in the product lifecycle. FDA:Guidance for Industry, Quality Systems Approach to Pharmaceutical CGMP Regulations 2006 QbD is an approach to rational design ; Building in quality from the development phase and throughout a product s life cycle QbD in conjunction with a quality system, provides a sound framework for the transfer of product knowledge and process understanding from drug development to the commercial manufacturing processes and for post-development changes and optimization. PROCESS & ANALYTICAL DEVELOPMENT PROCESS & ASSAY IMPLEMENTATION COMMERCIAL MANUFACTURING & LIFE CYCLE MANAGEMENT PDA Technical Report #60 Lifecycle approach to Process Validation Stage 1 Process Validation: Process Design Stage 2 Process Validation: Process Performance Qualification Stage 3 Process Validation: Continued Process Verification Commercial Product PRODUCT DISCONTINUATION 6

7 Product Learning and Improvement Process Learning and Improvement Continuous Learning and Improvement Product / Process Development & Lifecycle Management is an Iterative Process! PRODUCT DEVELOPMENT The development work flow cycles through the Execution Stages with each phase of clinical development Clinical data feed back into product development and ultimate definition of product s design space Knowledge gained from clinical manufacturing of the clinical lots can be leveraged for subsequent rounds of process development. As the development progresses, the total quality risk is reduced until acceptable for commercial licensure and manufacturing. Continuous process monitoring and verification enables process improvement over time. Discovery and Research Preclinical Research Target Product Profile Quality Target Product Profile Critical Quality Attributes PROCESS & ANALYTICAL DEVELOPMENT Parameter Impact Assessments Process Development Analytical Development Process Characterization Parameter Classification Process & Analytical Control Strategy Product Specification PROCESS & ASSAY IMPLEMENTATION Process Transfer Assay Transfer Master Batch Records Commissioning Equipment Validation Overall Manufacturing Control Strategy Process Qualification CLINICAL MANUFACTURING CLINICAL TRIALS Stage 1 Process Validation: Process Design Stage 2 Process Validation: Process Performance Qualification COMMERCIAL MANUFACTURING & LIFE CYCLE MANAGEMENT Instrument Calibration and Preventative Maintenance Batch Record Maintenance Alarm System Management Deviations Management Process Monitoring Process Requalification Continuous Impovement Commercial Product Stage 3 Process Validation: Continued Process Verification 7

8 Process & Analytical Development Product Development QbD Workflow for Process and Analytical Development Drug Discovery & Research QTPP Construction CQA Assessment Analytical Development Work Flow CQA Assay Selection Process Development Work Flow Analytical Methods Risk Assessment Parameter Risk Assessment Analytical-side Work Flow Analytical Development Plan Assay Development and Characterization Studies Process Development Plan Scale-down Model Development Process-side Work Flow Assay Qualification Scale-down Model Qualification Assay Pre-validation Process Development and Characterization Studies Test MethodsSpecifications Parameter Classification Work Flows join to make Process & Analytical Control Strategy Process Control Strategy Construction Product Release Assays and Specifications Process and Assay Implementation 8

9 PROCESS DEVELOPMENT PRODUCT DEVELOPMENT QbD Process Development Workflow & Risk Assessments Drug Discovery & Research Quantitative Outputs & Risk Scores QRM QTPP Construction CQA Assessment S CQA QRM Parameter Risk Assessment Development / Validation Planning Scale-down Model Development / Qualific. Process Development and Characterization Target Ranges (TR) Response Ratios (RR) Proven Acceptable Ranges (PARs) QRM Parameter Classification CPPs Impact Potential (IP) Score Normal Control Bands (NCB) QRM Process Control Strategy Construction Param. Controllability (PC) Score Param. Detectability Factor (DF) Score O CQA Process & Analytical Control Strategy Report Product Release Specifications D CQA PROCESS & ASSAY IMPLEMENTATION 9

10 Overall Risk Assessment Total Risk is a function of: QRM Severity of the event if it were to occur (S) Probability or Frequency of Occurrence (O) Probability of Detection (D) CQA Detection Risk Process Risk (Capability) Medical Severity Risk to Patient = S CQA x O CQA x D CQA 10

11 CQA Detection Risk Overall Risk Estimation for each CQA Contributing Components QRM Risk to Patient = S CQA x Medical Severity O CQA x D CQA Quality Attribute Assessment Score (1 10) Parameter Impact Potential (IP) Process Capability Cumulative Risk from Individual Parameters / Unit Operations (Rescaled 1-10) Critical Process Parameter Critical Process n Parameter Critical Process n Parameter Critical Process n Parameter n Parameter Risk = IP x PC x DF Parameter Controllability (PC) Parameter Detection Factor (DF) Derived from Analytical Testing Strategy (1 10) 11

12 Medical Severity Score S CQA from Quality Attribute Criticality Assessment Severity Scale applied to all three Clinical-effect Categories Efficacy S CQA = Max [S CQA ; S CQA ; S CQA ] PK Safety 10 9 Potency (Efficacy) Pharmaco Kinetics (PK) Safety & Immunigenicity 7 5 Critical Quality Attributes 3 1 Non-critical Quality Attributes Based on Severity (Patient Harm) only!, Independent of the probability of occurrence But weighted for Uncertainty.. 12

13 Severity Scale Severity Level Biological Activity (Potency) or Efficacy (Risk to Patient) Pharmacokinetics (PK)/ Pharmacodynamics (PD) Medical Severity S CQA Safety / Immunogenicity Immunogenicity Safety Very high NA Reduced potency for a chronic life-threatening disease is not considered a serious acute ( Very High impact.) event. Increased potency could be a serious safety event in case of an overdose. High Significant change (>40%) Reduced potency for a chronic life-threatening disease is considered High impact. NA Reduced PK/PD for a chronic life-threatening disease is not considered a serious acute ( Very High impact.) event. Moderate change (20-40%) with impact on PD Reduced potency for a chronic life-threatening disease is considered High impact. Hypersensitivity immune reaction Serious immunogenicity response Potentially serious impact to patient. Life-threatening or irreversible adverse events (AEs). Microbiology related infections. High impact to patient. Reversible AEs and/or loss of efficacy that is not lifethreatening Moderate Moderate change (20-40%) Moderate change (20-40%) with no impact on PD Low Acceptable change (<20%) Acceptable change (<20%) with no impact on PD Moderate immunogenicity Low immunogenicity potential Moderate impact to patient. AEs that can be managed by clinical treatment (i.e. dose titration, medication, etc.) No permanent harm. Low impact to patient. Safety or efficacy effect with minimal clinical significance. Temporary inconvenience or impairment. Very low No change No impact on PK or PD No effect of safety or efficacy. No patient harm. 13

14 Using Uncertainty to provide security margin to Severity Score Medical Severity S CQA Uncertainty-adjusted Severity Score: S adj = S Most Likely + U (S max -- S Most Likely ) Severity Scale Potency (Efficacy) PK Safety & Immunigenicity 10 9 Uncertainty Scores U = 9 (90%) S adj 7 S max = 7 5 S adj 3 U = 6 (60%) 1 S Most Likely = 3 U = 4 (40%) 14

15 Identifying Critical Quality Attributes (S > 4) QA-01 QA-02 QA-03 QA-04 QA-05 QA-06 QA-07 QA-08 QA-09 QA-10 QA-11 QA-12 QA-13 QA-14 QA-15 QA-16 QA-17 QA-18 QA-19 QA-20 QA-21 QA-22 QA-23 QA-24 QA-25 Attribute Ranking by Maximum Uncertainty-weighted Severity Score Non-CQA CQA (S adj ) CQA s QA-01 QA-02 QA-03 QA-04 QA-05 QA-06 QA-07 QA-08 QA-09 QA-10 QA-11 QA-12 QA-13 QA-14 QA-15 QA-16 QA-17 QA-18 QA-19 QA-20 QA-21 QA-22 QA-23 QA-24 QA-25 Medical Severity S CQA Attribute Ranking by Safety-related Only Uncertainty-weighted Severity Score

16 Using Quality Attribute Assessment Data to Prioritize Further Study QA-01 QA-02 QA-03 QA-04 QA-05 QA-06 QA-07 QA-08 QA-09 QA-10 QA-11 QA-12 QA-13 QA-14 QA-15 QA-16 QA-17 QA-18 QA-19 Ranking of Quality Attributes by Severity*Uncertainty Medical Severity S CQA Attribute Ranking by Maximum Uncertainty-weighted Severity Score (S most likely * U) Safety-related attributes with high severity but also highly certainty) (S adj ) Attributes indicated for further study (High combined severity and Uncertainty) 16

17 Performance Parameter (PP) or Critical Quality Attribute (CQA) Process Capability What is it? Process Capability O CQA Process capability is a measure of the process ability to maintain performance within specified limits Process Capability Index (if performance data available) Upper Specification Limit (USL) A Parameter can influence the magnitude of Process Output Variability by both: Output Lower Specification Limit Output (Attribute) Variability Process Performance Curve Input (Parameter) Variability The Slope of Process Performance Curve (i.e. the potential to impact the CQA or Performance Attribute) and. The degree of the parameter s variability (i.e. the ability to control the parameter) Input Operational Parameter (OP)

18 Process Capability Score O CQA Process Capability O CQA Estimated from cumulative risk over all Critical Process Parameters (CPPs) and Critical Material Attributes (CMAs) Over all CPPs and CMAs O CQA = S [IP Parameter i x PC Parameter i x DF Parameter i ] Parameter Impact Potential (IP): Degree of impact a parameter can have on the CQA if it were to vary over its characterized range Parameter Controllability (PC): Probability of deviating from its Control Range Parameter Detection Factor (DF): Risk of a deviation of the parameter from its normal operating range going undetected 18

19 Critical Quality Attribute (CQA1) Target Range CQA1 Response Range O CQA = S [IP i x PC i x DF i ] Parameter s Impact on CQA Determines it s Classification Impact is measured here as Response Ratio (Definition 1) Response Range of CQA1 to a large change in OP1 Process Performance Curve OP1 vs. CQA1 Target Range: A meaningful range in which the CQA is expected to be for a given unit operation or as a final specification Response Range: The range over which a CQA is observed to vary when a specific operating parameter is varied over the range of its study (characterized range). Response Ratio (Definition 1) for OP1 on CQA1 Response Range RR 1,1 = CQA1 Response Range for OP1 CQA1 Target Range Target Range CQA1 Characterized Range OP1 RR 1,1 > 0.2 Operating Parameter 1 (OP1) CPP Critical Process Parameter

20 Critical Quality Attribute (CQA1) Target Range CQA1 Mean Response Response Range Parameter s Impact on CQA Determines it s Classification Impact is measured here as Response Ratio (Definition 2) Response Range of CQA1 to a large change in OP1 Process Performance Curve OP1 vs. CQA1 CQA Mean Response (at Parameter set Point) Response Ratio (Definition 2) for OP1 on CQA1 Response Range RR 1,1 = O CQA = S [IP i x PC i x DF i ] Mean Response: The average response of a CQA observed for a given unit operation or as a final specification when process is run at setpoint Response Range: The range over which a CQA is observed to vary when a specific operating parameter is varied over the range of its study (characterized range). CQA1 Response Range for OP1 CQA1 Mean Response Mean Response Characterized Range OP1 RR 1,1 > 0.1 Set Point Operating Parameter 1 (OP1) CPP Critical Process Parameter

21 Definition of Critical Process Parameter (CPP). Should not take controllability into consideration.. O CQA = S [IP i x PC i x DF i ] Human Medicines Development and Evaluation EMA/430501/ August 2013 EMA-FDA pilot program for parallel assessment of Quality-by-Design applications: lessons learnt and Q&A resulting from the first parallel assessment In accordance with ICH Q8(R2) a critical process parameter is one whose variability has an impact on a critical quality attribute and therefore should be monitored or controlled to ensure the process produces the desired quality. The fact that a risk of failure is mitigated by applying a robust proactive control strategy should not allow for the underestimation of assigning criticality... in the Description of the Manufacturing Process and Process Controls and. Control of To Critical be as brave Steps as and the Intermediates people we help sections, the description of all parameters that have an impact on a CQA should be classified as critical. 21

22 Critical Quality Attribute (CQA1) Target Range CQA1 Response Range Determination of Parameter s Impact Potential (IP) Score IP Score is a measure of the degree of impact on the CQA if the parameter is varied of the extremes of its studied range. It is proportional to the Response Ratio (RR) if available. CQA1 Response Range for OP1 RR 1,1 = CQA1 Target Range Process Performance Curve OP1 vs. CQA1 Characterized Range OP1 Score Qualitative Criteria Response CQA highly sensitive to parameter fluctuation: Deviation of parameter or interacting parameter will lead to failure of DS or DP to meet quality targets; other steps cannot mitigate impact. Ratio 2.0 CQA very likely to be impacted with parameter fluctuation: 1.5 <2.0 Deviation of parameter or interacting parameter will likely lead to failure of DS or DP to meet quality targets; other steps unlikely to mitigate impact. CQA likely to be impacted with parameter fluctuation: Deviation of parameter or interacting parameter may 1.0 <1.5 lead to failure of DS or DP to meet quality targets; other steps may mitigate impact. CQA moderately impacted with parameter fluctuation: Deviation of parameter or interacting parameter 0.5 <1.0 unlikely to lead to failure of DS or DP to meet quality targets; or other steps will mitigate impact. CQA mildly affected by parameter or interacting parameter: Deviation of parameter or interacting parameter will not lead to failure of DS or DP to meet quality targets; mild parameter interactions and other steps will mitigate impact. O CQA = S [IP i x PC i x DF i ] 0.2 <0.5 CQA not affected by parameter or interacting parameter: <0.2 Deviation of parameter or interacting parameter will not lead to failure of DS/DP to meet quality targets; no entry point of failure. 22

23 Critical Quality Attribute (CQA1) Target Range CQA1 Response Range Determination of Proven Acceptable Range PAR is derived from Process Characterization Data O CQA = S [IP i x PC i x DF i ] Process Performance Curve OP1 vs. CQA1 CQA s Upper Specification Limit (USL) Set Point For example Shown. PAR bounded on upper side by causing the CQA to exceed its specified limit. PAR bounded on lower side by characterized range. The range studied did not go low enough to detect the lower limit of failure Characterized Range OP1 PAR Proven Acceptable Range (PAR) Operational Parameter (OP)

24 Concept of Parameter Controllability (PC) O CQA = S [IP i x PC i x DF i ] Operating Parameter Controllability is a function of both The accuracy to which a performance parameter needs to be controlled (i.e. Proven Acceptable Range) and.. The accuracy and precision that an operating parameter can practically be controlled (i.e. true equipment capability) Oscillation of Operating Parameter around Set Point Acceptable Operating Range (Proven Acceptable Range: PAR) Operational Parameter (OP) Set Point Proven Acceptable Range

25 Definition of a Normal Control Band (NCB) Normal Control Band (NCB) a new term. Represents the accuracy and precision that an operating parameter can practically be controlled (i.e. true equipment capability) It is independent of what range of control is required by the process O CQA = S [IP i x PC i x DF i ] Chosen specifically to distinguish it from a typical Normal Operating Range (NOR), which is typically ill-defined and often includes process considerations or needs Frequency Distribution for Operating Parameter Control around Set Point 1σ Normal Control Band NCB = +3σ -3σ +3σ Set Point NCB Operational Parameter (OP)

26 Parameter s NCB can be Compared to its PAR Probability of failure (parameter deviating outside of the PAR limits O CQA = S [IP i x PC i x DF i ] (Proven Acceptable Range: PAR) -3σ +3σ Failure Rate or Risk of Failure / Deviation Normal Control Band (NCB) Set Point PAR Operational Parameter (OP)

27 Calculation of Parameter Controllability Ratio (CR) Parameter Controllability Ratio is Indicative of Probability of excursion from Proven Acceptable Range (PAR) For a centered Range: CR CPP1 = For a non-centered Range: CR Lower = Lower NCB SP - PAR LL 3σ Lower PAR (Proven Acceptable Range: PAR) PAR 6σ = CR Upper = O CQA = S [IP i x PC i x DF i ] PAR NCB PAR LL - SP 3σ Upper PAR Upper NCB -3σ +3σ NCB Set Point PAR Operational Parameter (OP)

28 Performance Parameter (PP) or Critical Quality Attribute (CQA) O CQA = S [IP i x PC i x DF i ] Operating Parameter Controllability Ratio (PCR) is analogous to Process (Performance) Capability Index (Cp)! Process Performance Curve Upper Specification Limit Output CR Upper = PAR UL - SP 3σ Lower Specification Limit Failure Rate or Risk of Failure / Deviation Set Point Acceptable Operating Range NCB Operational Parameter (OP) Input

29 Performance Parameter (PP) or Critical Quality Attribute (CQA) Determination of Parameter Controllability (PC) Score O CQA = S [IP i x PC i x DF i ] PC Score is a measure of the probability of the parameter deviating far enough from its setpoint to exceed the limits of its Proven Acceptable Range (PAR). The score is proportional to the parameters controllability ratio if available. For a centered setpoint: Parameter s Controllability Ratio Width of Acceptable Operating Range = Width of Normal Control Band Process Performance Upper Limit Acceptable Control Range (PAR) Process Performance Lower Limit Normal Control Band (NCB) Process Performance Curve Set Point Score Qualitative Criteria Poor control of parameter: Parameter or any interacting parameters have >5% chance of deviating outside PARs. No downstream control of CQA. Poor to fair control of parameter: Parameter or any interacting parameters have >2% but 5% chance of deviating outside PARs. Downstream control of CQA unlikely or unreliable. Fair to normal control of parameter: Parameter or any interacting parameters have >0.3% but 2% chance of deviating outside PARs. Moderate control of parameter: Parameter or any interacting parameters have >0.01% but 0.3% chance of deviating outside PARs. Good to very good control of parameter: Parameter or any interacting parameters have >0.001% but 0.01% chance of deviating outside PARs. Excellent control of parameter: Parameter or any interacting parameters have 0.001% chance of deviating outside PARs. Controllability Ratio 0.83 > > > > >1.50 Operational Parameter (OP) 29

30 Determination of Parameter Detection Factor (DF) Applies to Redundant Measurement Device, not Controlling Device O CQA = S [IP i x PC i x DF i ] Detection Factor is the risk that a deviation from the control range will not be detected in time to allow corrective action. Does not apply to the primary control element. Alarm? TA Score 10 8 Qualitative Criteria No measurement in place; deviations in parameter cannot be detected without in-process testing. Immediate action is not possible. No direct measurement in place; deviations in parameter are not readily detected without in-process testing. Surrogate indications may be available and/or detection method has high probability of failure (false positive). Remedial action may be possible but is delayed or incomplete. TC TT TT Redundant Measuring Element Indirect ability to detect. Test not available but surrogate available, or delayed direct measurement prevents immediate action, or detection method has poor precision-to-tolerance and/or high failure rate. Direct detection. Measurement method moderate precision-to-tolerance and/or failure rate or delayed measurement results. Action may not be immediate. Direct measurement in place. Test has high precision-totolerance. Immediate reaction to deviations highly probable. 1 Ready detection of deviation with reliable on-line or offline real-time detection. Method has high precision-totolerance. Immediate reaction to deviations highly probable. 30

31 Determination of Overall Process Capability Score (O CQA ) O CQA an Estimation of Cumulative Risk over the process of CQA exceeding its specified limits O CQA Process Capability Assessment for a given CQA based on a cumulative score over all the parameter risk scores for individual CPPs/CMAs impacting the CQA. O CQA = S [IP Parameter i x PC Parameter i x DF Parameter i ] O CQA 31

32 Sub-scores for Individual Unit Operations can be used as Process Risk Profile Process Risk Score Unit Operation Composite Scores Process Capability O CQA Individual Parameter Scores UO1 Max Score Total Score Mean score CPP1 CPP2 CPP3 CMA1 Process UO2 CPP1 CPP2 CMA1 O CQA UO3 CPP1 CPP2 CPP3 CPP4 CPP3 CMA1 Estimation of Cumulative Risk UO4 CPP1 CPP2 CPP3 CMA1 UO5 CPP1 CPP2 CPP3 CMA1 UO6 CPP1 CPP2 CPP3 CMA1 32

33 Process & Analytical Control Strategy Construction Select First/ Next CQA Low-risk Parameter Medium-risk Parameter High-risk Parameter Failure Rate or Risk of Failure / Deviation NCB NCB Set Point 6σ 3σ < 3σ CPPs impacting CQA QRM Estimated Cumulative Failure Rate (Risk) Predicted Failure Rate (Risk) too High? no All CQAs Considered? no Add Control yes NCB Acceptable Operating Range Operational Parameter (OP) yes Control Strategy Complete

34 Critical Steps / Parameters may Require Additional Controls to Maintain all CQAs within Specifications. CPPs impacting CQA Select First/ Next CQA Estimated Cumulative Failure Rate (Risk) Predicted Failure Rate (Risk) too High? no All CQAs Considered? yes no Add Control yes QRM Control Strategy To be as brave Complete as the people we help Construction of Risk-based Control (A-Mab: A Case Study in Bioprocess Development. CMC Biotech Working Group) The level of control for each individual quality attribute is determined on the basis of the criticality level of the attribute and a risk assessment of the capability of the process to consistently deliver product that meets the acceptance criteria for each attribute. Based on this risk assessment results, a rational control strategy is formulated for each quality attribute by choosing the appropriate control elements. Thus it is the sum of the individual control strategies that represent the overall strategy Level of testing and controls is commensurate with risk. Risk is determined by the Criticality Level of the CQA, the process capability (or probability that a CQA would fail at a given step) and the probability of detection of a CQA failure. 34

35 Steps in Overall Construction of Control Strategy Pre-release Controls S CQA O CQA D CQA Step 1 Step 2 Process Capability Assessment and Individual Parameter Risk Mitigation Can high-risk CPPs control be improved? Each CQA yes Analytical Testing Strategy Design Parameters for further Monitoring Analytical Testing Plan Step 3 Step 4 Is residual risk to CQA acceptable? yes STOP Stability Monitoring Strategy Stability Program Post-release Controls Step 5 Process / Product Monitoring Strategy Process Monitoring Plan Step 6 Continued Process Verification Strategy CPV Protocol Control Strategy for CQA 35

36 Process Capability Assessment and Individual Parameter Risk Mitigation Assessment Mitigation Sequence done for each CPP/CMA for each Unit Operation for each CQA Step 1a Each CPP / CMA Calculate CPP/CMA s Initial Risk Score for CQA IP x PC x DF Step 1b, 1c Is CPP high risk for CQA? Can CPP s Control be Improved? Design/Add Improved Control to Baseline Process Parameters is Well Controlled. No further mitigation is required Step 1d All CPPs/CMAs Assessed For UO? Calculate Risk Score for Unit Operation Can CPP s Detection be Improved? Design/Add Improved Detection to Baseline Process For all CPPs / CMAs in Unit Operation For all Unit Operations in Process Calculate Risk Score for Process and Process Capability Score (O CQA ) O CQA Process Capability Assessment and Mitigation Complete for CQA 36

37 Assessments to Determine Analytical Testing Strategy Step 2a For Each CQA Calculate Overall Risk to Patient S CQA x O CQA x D CQA Step 2b Step 2c Is Overall Risk for CQA Acceptable? Propose Analytical Test and Process Location to Detect CQA D CQA Score Guidance 10: Validation Only 8: Interim Testing Only 6: Upstream In-process Test 4: Downstream In-process Test 2: Product Release Test Assess Test/Assay for Suitability to Detect CQA Document Testing Strategy and Rationale To be To as be brave as brave as the as people the people we help we help Analytical Testing Strategy Complete for CQA 37

38 Organization of CQAs in Control Strategy By Groups having Common Source and Controls Attribute Type Definition General Examples Product Protein Structure and Variants Any product protein structural characteristic or feature that may be classified as part of the product profile. Conformational features, including primary, secondary or tertiary structure of the protein amino acid backbone, Glycosylation and other common properties of glycoproteins:, Product-related impurities and degradants Drug Substance Process-Related Impurities and Residuals In contrast to Protein Product Structure and Variants, these attributes do not impart the desired biological function. Quality attributes may be associated with drug substance process-related reagents. Attributes may be associated with product-related impurities, including N- terminal or C-terminal truncations; proteolytically, chemically or thermally induced fragmentation/low molecular weight (LMW) species; high molecular weight (HMW) species such as aggregates, and chemically or thermally induced species such as deamidation, oxidation, as well as visible or subvisible particulates. Non-product related (foreign) particulates may also be considered. Chemical entities purposely added in unit operations of the drug substance manufacturing process. They are required for production but need to be removed from the final product. Other adventitious agents and potential process contaminants Attributes may be associated with adventitious agents and other potential contaminants. These attributes can appear in the final product if not detected and removed. These contaminants may be assumed likely, based on experience with similar processes and manufacturing history. Contaminating substances, particulates, and other chemical and/or biological entities may be inadvertently added to the product or otherwise find their way into the process., Living organisms can contaminate the process streams as adventitious agents such as bacteria, mycoplasma, molds, yeasts, and viruses or bacterial endotoxin., chemicals can also enter the process as contaminants of required raw materials or as leachates from process equipment and singleuse components, particulates can be introduced from process equipment and consumables by abrasion or other mechanical action. 38

39 Documentation of Process & Analytical Control Strategy Summary Table 1: Process & Analytical Control Strategy Summary by Critical Quality Attribute (CQA) Product Attribute Class And Critical Quality Attribute (CQA) Process Parameter (Input) Controls Control Risk # CPPS # RM Tests Analytical Testing Strategy IP Test 1.0 Product Molecule and Structural Variants Monitoring & CPV Terminal To Glycan be as brave 2 as the Low people we 9 help??? DS Release Test 1.1 Primary, Secondary and Higher-order Protein Structure DP Release test Amino Acid Sequence Low 3 2 Stability Testing CQA freq. CPP/ CMA Report Section Molecular CQA1 Low Molecular CQA2 Low Disulfide connectivity High Higher-order structure integrity High Glycosylation and Glycoforms Glyco-form 1 High Glyco-form 2 High 29 Antenal Distribution High 28 Terminal Glycan 1 High Impurities and Contaminants

40 Risk Assessment Methods Standardized by Building QRM Proto-type.. But Pain Points Limit its Scalability.. Excel work sheets Becoming large and cumbersome Not user friendly Revision history challenging Limited functionality Hard to maintain Specific Pain Points with Current QRM Approach Navigation and entering data rapidly is difficult due to limited field of data set visible at any one time. Building custom views of the data to suit all users rather; Currently a fixed format and sequence Sorting parameters by risk priority number without losing your place. Difficulty in storing or returning to previous versions or parameter sorts. Compress and/or customizing risk scores and heat maps from showing all parameters to a selected set of parameters, selected set of CQAs, just process steps, just unit operations or just the major unit procedures. Challenges with two-dimensional nature of Excel in showing multiple relationships (need to move towards a relational database approach) Integration with other systems (e.g. for capturing data, connection to process maps, process data) 40

41 User Requirements for Database-driven QRM Tool Organization and automation of various risk assessments in QbD work flow Ability to build custom views Rapid navigation and data entry data. Easy sorting of parameters by risk priority number. Ability to save and retrieve previous versions of parameter sorts Rapid compression of parameter list and heat maps by step, unit operation, sub-process Integrateable with other electronic systems and repositories Options to expand functionality at later date to include; Full FMEAs for Control Strategy QRM Material risk assessments Leachables & Extractables data base Root cause diagnosis 41

42 Deployment Strategy for QbD and Control Strategy Conclusions / Lessons Learned. So far Leadership and Management 1. Initiation required a critical mass of decision makers and a high-level program sponsor 2. Dedicated full-time program leader 3. High visibility and management support is key 4. Control strategy should be focal point as the culmination of QbD work 5. Engaged and knowledgeable team to develop methods and provide guidance 6. Ongoing management support needed to ensure universal and consistent adoption the QbD Czar! 42

43 Deployment Strategy for QbD and Control Strategy Conclusions and Lesson Learned - continued Project Management 1. Comprehensive strategy needed; Requires top down coordination; Project management support very helpful to maintain momentum 2. Challenge in estimating the level of effort required and scheduling of resources for QbD Team and ccmc Pilot team 3. Some ongoing confusion around scope 1. Early-phase or Commercial products and processes 2. Internal manufacturing or external (CMO) 3. Large molecule and small molecule? 4. Lack of clarity about phase-appropriateness of QbD expectations 43

44 Deployment Strategy for QbD and Control Strategy Conclusions and Lesson Learned - continued Change Management and Behavior 1. Change management needs initially underestimated and underresourced; Change management needs to be addressed early during implementation 2. QbD is an evolution. Changes and optimization of the business process and methodology should be expected. 3. Need to counter considerable angst by development people that they would be forced into a single, rigid, or overly quantitative way of doing things 4. Need to reinforce notion that overall task of designing, filing, and maintaining process over its lifecycle is simplified with the proper development and documentation done up front. Each step in the work flow generates information needed in later steps that may be conducted by other people/groups. 44

45 Deployment Strategy for QbD and Control Strategy Conclusions and Lesson Learned - continued 1. Common definitions and concepts critical; Alignment with Quality Systems very helpful as soon as possible. Ultimately required. 2. Alignment with a comprehensive risk-based manufacturing control strategy will be required to leverage QbD-based process understanding. 3. Clear document hierarchy, standardized templates helpful 4. Electronic systems required for successful wider deployment and data sharing. Ultimately systems integration will be required for: Product development and Pre-clinical data Process development data Systems Electronic document management 45

46 Next Steps for Shire Integration of QbD with commercialization business process Additional guidance on raw material assessment to be developed Align overall Manufacturing Control Strategy Required to realize full potential of QbD efforts Further integration/alignment with Quality Systems necessary Integration with Engineering design methods ultimately necessary Extend guidance into Process Validation / CPV Extend guidance to cover Analytical (aqbd) work flow Design and deploy database application to contain and assist QRM assessments 46

47 . 47