1. A Qualitative Decision Making Tool to Aid in Defining the Number of Lots for a Process Validation Campaign Leslie Sidor — Amgen Inc. 2012 Midwest Biopharmaceutical Statistics Workshop May 22, 2012

2. 2 Operational Excellence Outline  Where is Amgen today  Why a qualitative tool  Tool development  Conclusions

3. 3 Operational Excellence Commercial Production Post-approval Development PAI & Launch Prep Val. & File Prep Process Char. Comm. Process Development FIH Process Development Amgen Commercialization Process Validation Campaign P3 Campaign Phase 1 & 2 Campaigns Amgen Production Activity Lifecycle Approach to Process Commercialization Began 10 years Ago Magnitude of post approval changes can take process back into Commercialization Amgen’s old paradigm: Validation is a single event in time using 3 lots

4. 4 Operational Excellence Post- approval Development PAI & Launch Prep Conf. & File prep Process Char. Comm. Process Development FIH Process Development Amgen Commercialization Process Process Validation Scope Stage 1 Process Design Stage 2 PPQ Stage 3 Continued Process Verification (CPV) As defined in the FDA Process Validation Guidance (Jan 2011) Conf. Camp. P3 Campaign Phase 1 & 2 Campaigns Amgen Production Activity Commercial Production Lifecycle Approach to Process Validation Began 10 years Ago Magnitude of post approval changes can take process back into Commercialization Stage 2 will flex in size to assure “process is capable of reproducible commercial manufacturing”

5. 5 Operational Excellence Traditional Statistical Tools to assess process capability Require N > 30 lots  Control charts –Lot release –In process controls –Analytical method performance –Stability data  Process capability indexes –Lot release –In process controls Need to leverage qualitative information in Stage 1 to avoid statistical issues Inadequate sample size creates under or over reaction to signals

6. 6 Operational Excellence Goal: Leverage information from Stage 1 using a decision making tool to define size of PPQ Stage 1 Process Design Stage 2 PPQ Stage 3 CPV Stage 1 Design of Experiments Multiple Risk Assessments Critical Quality Attribute Matrix Comparability Etc. Amgen Activities Stage 2 PV protocols Validation report Product comparability Process comparability NC trending Etc. Stage 3 Control charting for numerous quantitative parameters Stability trending Process capability assessment Analytical method performance NC trending Quality System performance Etc. Integrate Stage 1 information + indirect performance of the facility using a Decision Making Tool

7. 7 Operational Excellence Attributes of the Decision Making Tool using an index  Tool maps qualitative factors into a numerical index to describe factors that impact ability to control the process  Tool uses qualitative characteristics of performance –Direct measures –Indirect measures  Tool needs to be… –Objective –Simple to use –Adjustable to allow for continuous improvement –Appropriately specific

8. 8 Operational Excellence Relative Ranking for Decision Making  Source: US Coast Guard, Field Demonstration Workshop on Performance-Based Inspection of Vessels Entering the St. Lawrence Seaway (Prioritizing Vessels for US Coast Guard Inspection), April 2000 For Internal Use Only. Amgen Confidential.

9. 9 Operational Excellence Step 1: Define what the index will represent  What does this index represent? –Is it consequence based: the final number represents the impact from consequences of interest. –Consequence of the index: description of the process by understanding sources of variability List of possible sources of variation that can impact process understanding Factors affecting the process understanding Scoring Factors Items of interest for scoring Scoring of the most impactful sources of variability for a molecule Scoring of Performance Characteristics that describe process understanding. Index will drive the size of the PPQ campaign

10. 10 Operational Excellence 2. Brainstorm factors that could affect the index  Team members: 16 staff over 8 functional areas  Over 100 factors were identified  Factors were classified into 8 Performance Characteristics  Factors were modified to remove ambiguous language (e.g.: robust, approximate, successful, etc.)  DISADVANTAGE: Miss a key functional area, factor list will be incomplete -Process -Product -Raw Materials -Facility -Risk -Regulatory -Analytical Methods -Documentation

11. 11 Operational Excellence Step 3: Identify specific situations for which specific actions are required QuestionAnswerAction Is this a new facility?YesProceed to PPQ, but Perform the maximum amount of work based on the action threshold Is process locked?NoDo not proceed to PPQ Is a PPQ protocol in place? NoDo not proceed to PPQ These questions are asked before the tool is used and overrides any decision from the tool

12. 12 Operational Excellence Steps 4 & 5: Select a basic scoring or indexing & develop scoring scales for each factor Performance Characteristic FactorNo: Less process understanding ProductProduct comparability has been demonstrated at commercial scale No =1 DP only: Experience with multiple DS ages Yes=0 ProcessProcess tracking active in Clinical manufacturing No=1 Raw MaterialsAll raw materials have been used for the same application at commercial scale No=1 Average SCORE75% Use a binary response and compute % in the “No’s”

13. 13 Operational Excellence Considerations for scoring  All factors are not created equal  Apply appropriate weighting: Should a “No” for some factors be worth 2 points?  Possibilities for average score –Compute straight % for the “NO’s” (assumes each factor = 1 point if NO) –Weight some factors greater than 1 point, then compute %

14. 14 Operational Excellence Step 6: Set action threshold for the index Score (% “No’s)# of Lots for PPQ + Transition plan into CPV > 80%> 6+ monitor PPQ parameters for next 15 lots; then move to CPV >60% to < 80 %5 to 6+ monitor PPQ parameters for next 10 lots; then move into CPV >40 % to < 60%3 to 4+ monitor PPQ parameters for next 5 lots; then move into CPV < 40%1 to 2Move into CPV Option 1: Define # of lots Option 2: Define # of lots + a transition plan No Decision has been made at Amgen to define the action threshold

15. 15 Operational Excellence Final steps are not complete  Step 7: Organize the scoring scale, index calculation and action thresholds into an SOP –A white paper will be written describing development and rationale –An SOP or training will be needed on how to use the tool –Documentation of the tool user’s answers and the justification of the answers will be incorporated into the tool  Step 8: Verify the documentation makes sense and refine as needed –Does the score make sense? –Did we miss some factors? –Does the weighting need further refinement?

16. 16 Operational Excellence Conclusions  Statistical tools are not appropriate for low sample sizes  Implement a qualitative decision making tool that leverages Stage 1 information –Follow systematic approach to develop tool –Test tool to look for gaps before implementation –Evaluate documentation for ease of use  Overall size of PPQ campaign also needs to consider comparability and stability requirements

17. 17 Operational Excellence References  Guidance for Industry Process Validation: General Principles and Practices, January 2011  Levy, P.: Determining and Justifying the Number of Process Validation Batches: Making Initial Batch Release Decision. Presented at ISPE: Lessons Learned from 483s Process Validation Track, February 2012  Field Demonstration Workshop on Performance-Based Inspection of Vessels Entering the St. Lawrence Seaway (Prioritizing Vessels for US Coast Guard Inspection); April 2000

18. 18 Operational Excellence  ABSTRACT:  The new FDA Process Validation guidance recommends a lifecycle approach for process validation by allowing the manufacturer to incorporate process and product understanding into the timing and size of the validation campaign (referred to in the FDA document as Process Performance Qualification or PPQ). This paper describes a qualitative decision making tool to help determine an appropriate size of the PPQ. The tool is based on criteria from seven key focus areas that aid in the understanding of process capability and product consistency. These seven areas are product understanding, process understanding, facility information, raw materials, status of analytical methods, regulatory considerations and level of documentation. The tool lists performance characteristics for each focus area. The user then lists a binary response of “yes” or “no” for each characteristic. The justification for the number of lots for a process validation campaign is then determined based on the percentage of “no” responses.