Regulatory and Strategic Considerations: What is CPV? (contd) Industry case studies provided highly detailed illustrations of how CPV is designed and

Similar documents
Principal approach to CPV :

Wrap Up Summary CMC Strategy Forum. Bridging Analytical Methods Jan 27, 2014

Control strategy and validation. Emanuela Lacana PhD Office of Biotechnology Products CDER/FDA

A COMPARATIVE FRAMEWORK BETWEEN NEW PRODUCT & LEGACY PRODUCT PROCESS VALIDATION

Themes & Key Points The Problem Statement

QUALITY ASSESSMENT METHODS FOR NEW PRODUCT LAUNCHES: PROCESS VALIDATION LIFECYCLE

Dr. Earl Dye CMC/GMP Considerations for Expedited Development Programs

Demonstrating a High Degree of Assurance in Stage 2 of the Process Validation Lifecycle

Advantages, Opportunities & Challenges of Adopting the Post Approval Change Management Plan (PACMP)

Process Capability: Practical Challenges to Implementation

PPQ-to-Approval Timelines Acceleration Approaches at BMS

A holistic regulatory approach to accelerated CMC development

Application of ICH Q12 Tools and Enablers

Product, Process Knowledge & SPC: PV Lifecycle Approach IFPAC January 2016, Arlington, VA

Established Conditions: Reportable CMC Changes for Approved Drug and Biologic Products Guidance for Industry

CMC Activities for Commercialization: PAI, QMS, and BLA/NDA. NEPDA Dinner Meeting Cambridge, MA 16 May 2018

An Industry Perspective: The Complexity of Postapproval CMC Changes and Proposed Regulatory Strategies. SPEAKER: Suzanne Murray Biogen

THE PACMP STRATEGY. March 13-16, 2018

Process Monitoring Applying QbD Principles in a Biopharmaceutical Environment

PROCESS VALIDATION ROCHAPON WACHAROTAYANKUN, PH.D.

Process Validation Guidelines. Report highlights 23 rd February 2018

Post-approval Change Management Protocols - Current Status and Next Steps on the Way towards a Global Tool

The FDA Just Arrived... Are You Ready? Presented By Sandra Lueken Sr. Director, Quality AstraZeneca Biologics

EVOLVING TRENDS IN THE USE OF STATISTICS FOR PROCESS VALIDATION IVT 3 RD ANNUAL STATISTICS IN VALIDATION JUNE 20-22, 2017

Common Issues in Qualification and Validation of Analytical Procedures

Guidance for Industry

Process Validation Lifecycle Approach: A Return to Science

ICH Q8/Q8(R)

FDA Quality Metrics Guidance

While the recognition

Challenges Integrating Regulatory Filings and Pre-Approval Inspection with the Expectations of Current Regulatory Guidance

EFFECTIVE MANAGEMENT OF THE PROCESS VALIDATION LIFECYCLE VALIDATION MASTER PLAN DEVELOPMENT

ICH Q12 Perspectives: The Robust PQS

Regulatory Updates for Biopharmaceutical Products:FDA Perspective

Post Approval Changes & Product Lifecycle Management

Clinical qualification of specifications - a Regulator s view

Regulatory Perspective on Analytical Method Validation During Product Development

4 Key Considerations When Engaging A New GMP Contract Service Provider

Strategic Implantation of PAT : FDA Perspective

Addressing the Paradigm Shift in Regulatory Inspections

Quality by Design Considerations for Analytical Procedures and Process Control

Regulatory perspective on setting clinically relevant specifications. Joslyn Brunelle, PhD Team Leader Office of Biotechnology Products

PhRMA REPS MARY OATES AND MOHEB NASR (RAPPORTEUR) ON ICH Q12

An Industry Perspective: Reducing the Complexity and Impact of Regulatory Changes in Latin America

Current Hotspots during CMC Evaluation a European Regulatory Perspective

CMC Forum Europe, 2013

PMDA Perspective: Regulatory Updates on Process Validation Standard

Process validation in medical devices

Process validation in medical devices

A Potential Innovative CMC Solution: Responding To Public Health Needs With An Accelerated Clinical Pathway A Vaccine Example

A Framework and Case Study for Implementing the New Process Validation Guidance

ICH Quality Implementation Working Group POINTS TO CONSIDER

PROCESS VALIDATION ANSM 2015 FDA 2011

Regulatory Compliance Inspections at MAH Offices: Implications for Manufacturers

Evolution of Quality Assessments Recent Trends in FDA Queries. Mike Saleh, Pfizer Inc.

Statistics in Validation. Tara Scherder CSO Supply, Arlenda, Inc

GenoLogics LIMS Helps EdgeBio Build Services Business

Appropriate Control Strategies Eliminate the Need for Redundant Testing of Pharmaceutical Products

WHITE PAPER. Analytics Software. Find what Matters

FDA s Draft Guidance Request for Quality Metrics: What All Drug and Biologics Manufacturers Need to Know June 2017

QbD and the New Process Validation Guidance

A Late-Stage Monoclonal Antibody in the FDA QbD Pilot Program: Moving from Concepts to Implementation

Quality by Design (QbD)

Regulation of Active Pharmaceutical Ingredients (API)

Challenges with Establishing a Control Strategy for Biosimilars

Synopsis: FDA Process Validation Guidance

ICH Q12 Technical and Regulatory Considerations for Pharmaceutical Product Lifecycle Management

Planning of Application

7 STEPS TO SUCCESSFUL RETENTION AUTOMATION YOUR GUIDE TO MAXIMIZING REVENUE FROM YOUR CUSTOMER DATA

HOW YOUR CAREER BACKGROUND CAN HELP YOU BECOME A BUSINESS ANALYST

Application of ICH Q12 Tools and Enablers Post-Approval Lifecycle Management Protocols

Replacing Analytical Methods for Release and Stability Testing CBER Perspective

7 STEPS TO SUCCESSFUL RETENTION AUTOMATION YOUR GUIDE TO MAXIMIZING REVENUE FROM YOUR CUSTOMER DATA

PPTA Regulatory Workshop June 13, 2016

ICH Q12 (Pharmaceutical Product Lifecycle Management): PMDA Perspective

Overview of Statistics used in QbD Throughout the Product Lifecycle

Challenges in Implementing Knowledge Management within ICH Q10

The Ever-increasing Complexity of Biotech Changes - A Pledge for Global Convergence

Optimising the management of post-approval changes for patients timely access to medicines

Risk-Based Approach to SAS Program Validation

ICH Q10/WHO TRS A Regulatory Perspective

Lifecycle Product Quality Risk Management

Demystify Governance, Risk & Compliance For Lifesciences

PROCESS VALIDATION Jaap Koster. Pharmaceutical Consultancy Services, All rights reserved.

ANALYTICAL VALIDATION CHALLENGES DURING THE RAPID DEVELOPMENT OF KEYTRUDA

Use in Process Characterization. Nathan McKnight

Introduction to Control Charts

Current Features of USFDA and EMA Process Validation Guidance

ICH Q12 : A Unique Opportunity to Realize 21st Century Quality Vision

Implementation of Lifecycle Validation Practices at CMOs 1

Harmonizing Standards and Specifications

The Use of Pharmaceutical Stability Tools in Medical Device Stability Programs

5 ESSENTIAL MARKETING AUTOMATION CAMPAIGNS

Quality by Design: An Attempt to Jumpstart. Peter Calcott, Ph.D. President, Calcott Consulting

Linking Regulatory Commitments to Post Approval Changes Robert Iser

Post Approval Change Management Protocols/Comparability Protocols (PACMPs/CPs) Current and Future State

Harmonizing Manufacturing across Multiple Sites and Regions

FDA Quality Metrics, Data Integrity and Application of Statistics Throughout Process Validation in a Global Economy

Industry Perspective on Manufacturing in Early Development

Cell Therapy Product Manufacturing Considerations. July 17, 2017 CMC Strategy Forum Mo Heidaran, Ph.D.

Transcription:

Regulatory and Strategic Considerations: What is CPV? FDA affirmed we share the goal of safe, effective products from controlled, efficient processes. They gave comprehensive overview of the regulatory concepts for Total Control Strategy (FDA 2011 PV guidance; ICH Q6B, Q8, Q10, Q11, upcoming Q12) and statutory requirements when making changes to an approved control strategy (21CFR601.12). No individual element works alone; it is a combination of sound process characterization linked to sound product characterization based on collecting sufficient data, monitoring performance trends, evaluating causes/effects of variations, applying appropriate statistical models and conducting risk assessments to understand and optimize the process/product control strategy. CPV formalizes a firm s procedural, logistical, mechanistic expectations for state of control What is NOT CPV? Frequent ongoing process changes that are outside of what was in BLA to tweak improvements (CPV is not a replacement for sound process development work) Changes in process steps that alter product quality beyond clinical ranges (can t unilaterally adjust approved specs to fit product from a modified process) New draft guidance: Reportable CMC Changes for Approved Products Clarifies per ICH CTD section where established conditions exist for DS and DP when the application is approved Gives a tangible overview of exactly which elements comprise the total control strategy as defined in the relevant sections of the dossier for DS and DP Why wasn t much mentioned about stability in the CPV control strategy? FDA forgot to put it on the slide A-Mab team debated whether to include stability or not. Need to apply it to entire process (DS and DP); stability is integral to the overall control strategy. Definition of process is the product ; for CPV to be effective it should go all the way to the point where the last DS bottle gets form/filled into the last DP batch, then the last patient gets the last DP dose on the last day of shelf life

Regulatory and Strategic Considerations: What is CPV? (contd) Industry case studies provided highly detailed illustrations of how CPV is designed and managed in different firms. Information from CPV strategies provide justification/rationale for where process control elements can be better refined. The CPV plan should provide a prioritized subset of critical process outputs (at a minimum). Baxalta walked through a major generic case study: BPOG CPV whitepaper (designed with similar approach as the A-Mab case study; multiple contributors and combined best practices) CPV is a logical consequence of as product development following QbD principles; differences can be recognised and accounted for, quickly establishing a consistent position (or else clear reasons for divergence!). NovoNordisk gave very detailed illustration of their approach: CPV is designed to meet three goals (1) Maintain validated state of the product, process and system, (2) Enable continuous improvement (3) Meet regulatory requirement for lifecycle validation. CPV strategy includes comprehensive review, documentation, evaluation impact of changes (science-based, data driven); Based on annual verification exercises (calendar cycle) per product per production unit; Involves managers, operations, SMEs and QA for each product/system go through ALL inputs and outputs from their systems that impact the process/product. Deliverable VSS (Validation Status Summary) shared with management for organizational visibility via quality management review; includes issues, conclusions and action items (with plan and timelines) Learnings: Establishing a CPV plan requires a cross-functional highly knowledgeable team to deliver real results; a coordinated CPV program can increase efficiency of continuous improvement activities; CPV implementation needs good IT database management. A sound CPV should address these questions: Is the process in a state of control? Is it as efficient as it can be? Are risks reduced to acceptable limits?

Practical and Statistical Considerations: How should signals from CPV interface with your deviation systems? What can/should you be able to do with the outputs of CPV? FDA guideline on expedited programs says PV could be flexed if necessary for lifesaving products why cant it be used for all products? There are no hard rules on what elements are flexed, but even if it is a BT product there still needs to be sufficient data to show the mfr process is able to produce the materials consistently Why can t we just ditch the PV 3 lot scheme and make CPV the only element required, and have it be assessed only during PAI? FDA: 3 lots is a negative experiment; if you can t do it, it says a lot; if you can, it doesn t say much. Failing the ability to make 3 consecutive passing lots is highly telling about the state of process control and your actual readiness for commercial production. FDA needs data that shows the process can do what it is purported to do at LEAST three times in a row. Can the CPV include the 3 lot scheme that is inspected, not reviewed? NO ONE can review and approve a process without ANY relevant data that can support the process can function to make the intended complex biologically-derived product FDA does not always do PAI for a new product and does not always look at the entire process control plan at the level of a reviewer looks at PPQ in a dossier; PAI is a very short snapshot of the site capabilities that does not allow time to review CPV plan in depth

Practical and Statistical Considerations: How should signals from CPV interface with your deviation systems? What can/should you be able to do with the outputs of CPV? Discussions on Established Conditions (new FDA guidance); are we making progress or are we still putting things in boxes? FDA is supporting lifecycle approaches within the expectations of data to support established clinical ranges for product safety and efficacy and assure conformance to approve specifications. How much (if any) of the CPV can be pre-approved for a sponsor to work within for licensed products? The FDA would see the control strategy defined and justified in the dossier, but the CPV post-approval internal plan on inspection. Health Canada had a CPV plan in dossier with prospective decision-making steps that was approved. Maybe some strategic elements could be included in the dossier that could be discussed during the review cycle? FDA possibly; depends on what details are specifically included for them to review. Health Canada had a CPV plan in dossier with prospective decision-making steps that was approved. Maybe some strategic elements could be included in the dossier that could be discussed during the review cycle? FDA possibly; depends on what details are specifically included for them to review. Are there situations where you could make a change and notify agency after the fact? There are cases where a prospective change plan can be approved in the dossier, like dropping HCP or HC DNA after X lots are within spec limits; these plans would be approved as a part of the product application

Regulatory and Strategic Considerations: How does CPV fit into the control strategy lifecycle? Lots of discussions on who is the intended audience for CPV plans: Internal management tool or external regulatory document (either in dossier or during inspection)? An internal document, not a regulatory dossier element but is provided upon inspection; they are alerts, not deviations. Differences in regulatory histories of biotech companies lead to different perspectives on response(s) to regulatory guidelines Not all companies agree that the CPV must be a regulatory commitment, some want to keep it as an internal best practices then use the outcomes to communicate to regulators plans for defined improvements (via PAC comparability studies). How much (if any) of the CPV can be pre-approved for a sponsor to work within for licensed products? The FDA would see the control strategy defined and justified in the dossier, but the CPV post-approval internal plan on inspection. Maybe some strategic elements could be included in the dossier that could be discussed during the review cycle? FDA possibly; depends on what details are specifically included for them to review. Health Canada had a CPV PALM plan in dossier with prospective, decision making steps that was approved with sufficient detail and supportive data to allow HC to evaluate. Major topic: How to implement process changes arising from CPV exercises? What is within the approved design space / control strategy vs what changes are outside of those regulatory guardbands? Is there a possible reporting paradigm that could serve the principles of CPV as applied to complex biological processes/products and still meet the statutory requirements of regulatory authorities for approving products? How does CPV relate to ICHQ12? The EWP group is struggling with what elements would be binding vs non binding ie what is managed in the PQS and what is in the dossier; even the terms plan vs strategy are debated.

Practical and Statistical Considerations: How do you capture experience from long term process performance and integrate this not only into limits for process control but also specs? Excellent presentations on the various approaches to design and implement control charts for monitoring process, product and (thanks, Brian!) method true performance capabilities. Especially with legacy products where you are drinking from a fire hose. Many examples of what the data can tell you, and how to reasonably interpret what it is truly telling you. [note to file: no more red lines, red dots]. Data trends can falsely enhanced or masked if all you are looking at is what you expect to see (gorilla is ignored). Amgen gave superb tutorial on pros vs cons of using Cpk (potential capabilities) vs Ppk (actual capabilities) to assess statistical process control (a operational state in which only common cause variation is evident). Gave great examples of common cause, special cause, and long term common cause data sets. It is vital to understand the nature of the underlying data for interpreting Cpk or Ppk. Biogen pointed out that the goal should not be to set limits that become fossilized, but to set limits supported with initial data, then refine them as future data are collected and analyzed. You should constantly review performance trends not only to confirm that the system is staying in control but to learn where you can continuously improve efficiency and robustness. (Concepts in FDA 2011 guidance can be applied to analytical method validation lifecycle). To do this requires good centralized data management (more than just a LIMS repository). Discoverant is a validatable application that integrates data from multiple sources for analysis and reporting. Generates a wide array of analysis outputs with alerts; produces trending reports. Other applications: stability data, method invalid rates.

Regulatory and Strategic Considerations: What is the relationship between enhanced development (QbD approaches) and CPV? MedImmune provided good granularity on their perspectives on line of sight elements of coordinated control strategy for product lifecycle: Development studies (characterization and stability) data with (limited) commercial batch data inform initial limits in the control strategy. From there, monitor routine product lifecycle events (robustness) and apply risk based and appropriate statistical modeling to lifecycle management of product quality and supply. Lilly also presented their perspectives on how they use the 3 stage strategy for process characterization, process validation, and continuous process control (verification). Starts with applicable prior knowledge then builds the process specific elements with staged risk assessment gates. PPQ system includes more than just 3-5 lots; various unit operation assessment protocols plus process comparability protocols, ending with a decision gate for go/no go to PV. CPV is captured in a prospective protocol with real-time assessment of excursions/performance trends and APR (Need a good suite of sophisticated electronic data collation and analysis systems) Discussions: It may require 30-50 batches to get statistically meaningful information; Bayesian approaches could leverage limited development data and model the predicted path for process; Might be able to use small scale models to generate more data sets, but must demonstrate small scale model is comparable to the full scale process if data are to be relevant CPV might indicate a control has to be ADDED rather than omitted; if a new variable arises that was not recognized before.

Regulatory and Strategic Considerations: What is the relationship between enhanced development (QbD approaches) and CPV? Baxalta provided insight on why biotech processes are not normal (phrasing! Violations of Normality in BioPharm ); gave examples of how standard models should be adapted for biopharm operational monitoring. Asserted that QBD elements (meaningful data, process understanding, risk assessment) are not only nice but are necessary to generating relevant statistical evaluations. The goal should be to obtain relevant signals not just statistical ones. Gave details on how QBD input feeds into CPV plans. Echoed FDA: CPV is not a change control system, a replacement of OOS investigation, or new CAPA procedure. Pfizer case studies with highly diverse molecules (vaccine, Mab, ADC) but common strategic approaches; for each, invests resources in early process/product understanding, leverages platform knowledge and prior experiences with each type of product to guide process and analytical decisions; goal is to build control strategy to enable commercialization with postapproval CPV. Lessons learned: Leverage prior experiences in guiding early process/product choices; feedback from regulators is important; can make business case for QBD and CPV, with sufficient data can get approval to remove from some tests before 30 lots eg HCP and DNA, but saw different opinions among regulators on when and how to tighten specs after approval With CPV, has industry seen any issues that have ADDED a control element that wasn t there before? One has had this experience mostly due to upstream elements; others have not yet but pay close attention to raw materials as high risk points. Analytical method changes have revealed previously unknown species that then triggered monitoring to confirm/refute criticality

Regulatory and Strategic Considerations: How can CPV be applied to legacy products? GSK showed how they utilize CPV data sets with over 30 commercial products with data from 2006-2015 (over 1000 parameters) from multiple production sites. CPV analyses harmonized by use of validated in-house software automatized analysis with long-term storage and reliability. Decisions are driven by process stability and capability, investigation of special causes, and defined process performance levels (good, acceptable, poor). Gave great overview of change point control analyses models and how to best represent complex data sets to effectively communication in trend team meetings. Merck and Lilly (oldest biologics!) showed how they assess legacy products for CPV principles. Legacy products have enough data to assess what is normal and what is not. Do not assume early PD can be fully comprehensive; No guarantee against new sources of variability; noncqa/noncpp data are essential for problem-solving; fix all fixable issues. Most frequently reported source of unexpected variations after PPQ for commercial products: RAW MATERIALS. Magical 30 commercial lots is it enough to really show true operational variances? Or too much? ADC example 30 lots might not be enough to tighten specs to reflect real process variations. For a known source of variance (eg process residual), 30 lots might be overkill. For an unknown type or source of variance, would be based on statistical prediction interval. So might need more than 30 lots to increase rigor of the intervals. Can we investigate false positive too early? Risk based decision but is there a ROT for # batches to avoid tracking white rabbits? Look at all data have SMEs make reasonable choices about what you see; don t over or under react. Hindsight caution from audience: Do not over react to the first trend violations

Regulatory and Strategic Considerations: How can CPV be applied to legacy products? Updating legacy control strategies with upgraded or downgraded elements: Adding is not a problem with regulators (sure, go ahead!, but removing an approved control element would require sufficient data to justify the change and support the impact to product. In some cases FDA could approve the removal of a parameter if the data sets are adequately abundant and show good history of control. Could also move control points upstream or downstream (eg raw materials) to more effectively monitor process consistency. When does CPP failure cross over to quality system? If it fails CQA spec, immediately. If it is an excursion but not limit failure, depends on how close to the limit, if seeing a pattern of CPP issues, would investigate to see if there is something going amiss. Control chart excursion signals how do firms handle them within the quality system? Upcoming BPOQ paper will cover what are appropriate types of responses to alerts (tiered level?) that are not OOS under the PQS (including documentation of the decisions with justifications) Enhanced monitoring of purposeful change vs detection of change followed by monitoring? Statistics are not different but response may be. Bigger deal than normal but not big enough to redo PV; might include short-term enhanced monitoring or increased sampling to confirm prediction PAT/CPV has improved the process of problem solving across sites when one experiences a problem; one site called another to say pay attention to something that was causing them problems; really shortens the diagnostics and investigations (may not shorten the remediation time/efforts); saved many batches from possible problems by sharing knowledge of process control. Cross-site discussions help flag trends/issues that might be missed if each site only looked inward.

Practical and Statistical Considerations: What can/should you be able to do with the outputs of CPV? Q from FDA: How do sponsors envision interacting with FDA on their CPV plans? What information do they WANT to share vs prefer to keep internally for monitoring and decision making? If applicant is willing to share eg in supplement the CPV plans then hopefully the agency will review it and provide comment. Isn t there an initiative to have cross training of inspectors and reviewers? Both CDER and CBER have multidisciplinary teams for review/inspection. But, in the field time is quite constrained so if CPV is only looked at there it would not get adequate attention. Integration of review/inspection activities for BLA vs PAS vs inspection still evolving; FDA review office does communicate with inspection on initial and follow up items; goal is cross training but there will always have to be SMEs because no one person can know everything at sufficient depth to critically evaluate it. What about global experiences with CPV (more than FDA and HC)? Not sure how progressive other regulatory agencies are because there is limited open dialog or transparency with other regions. More publications, and more platform/meeting discussions (Heads-Up: CMC Forum Global Topic 2016?), would be very useful to continue to open interactions (industry case studies for mutual education on possible approaches).

Practical and Statistical Considerations: What do you do in the early stages of commercial production just after PPQ? Should your response signals depend on how much experience you have? Amgen showed that although small data sets are very challenging but can be leveraged with appropriate models. PD is accumulation of data to understand the process to be validated, but most biotech companies don t do enough PD to really understand the process control strategy (scientific or business). So the control strategy would be expected to evolve as more data are collected. Goal is to manage all sources of variability using CPV data monitoring. How do we inform regulators of the outcomes of trend analyses, eg tightening or adding an internal control over raw materials? Current paradigm: If that is a change within your approved control strategy you should be able to implement it and notify them ie by Annual report. If you find something that is in an established condition but requires alterations then it must be reported; level of reporting should be based on the risk of the change. Emerging paradigm? ICHQ12discusses this as a possible different notification category. CMOs might be able to leverage collated data from platform processes used with multiple sponsors, but would require data to be shared with sponsors so would have to be blinded. Even if platformed, each process and product is different; cannot directly extrapolate features and controls across products (different patient populations, dosages, ROA, etc ) so each has to be reviewed on a case by case basis. Has anyone worked with CMOs on CPV issues? Crickets.. Have sponsors seen CPV programs add value to project teams; is cost/benefit worth it? Yes, if the data sets are electronic, harmonized content and formats; information is more widely and more readily available, enhancing rapid communications among SMEs to allow real-time reactions to issues (rather than protracted cumbersome internal processes).

Practical and Statistical Considerations: The system requires quite a sophisticated combination of process and statistical knowledge. How do you help internal and external auditors to understand that? Elegant presentation formats for statistical analyses are not a trivial issue the goal is effective communication to an audience that is by no means ignorant but not necessarily experts in mathematical transformations of complex data. Must find graphical means to clearly and most accurately - illustrate complex analyses containing relevant statistical annotations (eg CI, TI, etc). SPC control charts (rolling ranges), surface plots, rising sun plots. If nobody can follow what you are doing, the practical value will be lost to your organization. Various illustrations of software packages (see speaker slides for details; alas, some are internal). But cannot simply plug and play for automated data management; must have SMEs engaged in what goes in so that what comes out is meaningful. At least two 483 issued on inappropriate use of statistics to make (incorrect) decisions on process control ** JO: Let s start publishing more examples of how to use these tools correctly for biotech process/products to inform the entire field ** How much computer validation is typical for data management software? Main drivers are data integrity and data accuracy so that internal decisions can be made on reliable, retrievable, unadulterated data. Manual data review in one firm is done on tiered level of control based on how the data are to be used; could apply the same design to electronic systems; BPOQ group is working on this issue now for publication soon. FDA - Any part of the system that is in the cgmp system would require Part 11 level validation (predicate records).

Practical and Statistical Considerations: What do you do in the early stages of commercial production just after PPQ? Should your response signals depend on how much experience you have? Has a robust process design space been filed, and what did it buy the sponsor? FDA review office has seen and approved one for US/DS operations; it included what variances do and do not require reporting. HC has seen one for process design space but it never got approved (firm withdrew it) QbD pilot program had some quite good submissions but products didn t meet clinical trials so were dropped; FDA is seeing much more QBD like approaches in filings. FDA supports QBD approaches and would like to see actual examples of how CPV can synergistically support the design space that was modeled and predicted. Maybe it turns out that the amount of work needed to get the claim for robust design space would be far greater than just making the changes and doing typical comparability. One speaker said they are actually generating a huge amount of design space data but are not putting it into a filing yet (it is used for FIO internally) How does FDA handle newly-discovered entities that were previously present in legacy products? PS80 screws up endotoxin assays so there are plenty of approved applications with faulty endo tests (most sponsors are actively addressing them though ) Newly discovered species revealed by more sensitive/specific analytical methods are possible, so sponsors should collect data, assess risks, and engage in discussions with regulators on appropriate paths forward with existing or adjusted specifications. FDA does not want to penalize sponsors for upgrading methods. New paradigm at FDA: We don t have enough resources to evaluate everything everywhere. Riskbased approach to inspection will include evaluation of firms quality risk management systems suitability for maintaining a suitable state of control to generate products of intended quality

ADDITIONAL SUMMARY NOTES FROM TIM and JULIA (PROVIDED AFTER THE FORUM) The success of a strategic continued process verification plan is founded on scientifically meaningful specifications. When limits are set based on batch release data there are limited opportunities to monitor and react to the manufacturing process. Many of the speakers pointed out the complications of a CPV plan when there are shifts in the process, perhaps due to campaign effects. It was highlighted that usual SPC rules tend to fail in these cases and that perhaps a more strategic approach looking for shifts in campaigns (or special cause variability) and shifts within campaigns (common cause variability) be used. Statistical approaches which have been used in other applications such as genomics were illustrated. With regard to statistics versus technical, a point to note is the utility of partnership, and the value in including statisticians on strategic oversight teams. Clearly there was a message of KISS but that goes two ways. Process and analytical folks have to clearly describe their goals (what s the question) and the complexity of their processes to make optimal use of statistical analysis. Emphasis was on applying statistics thoughtfully within the context of CPV, to gain the most power identifying real signals of change, and fitting meaningful models to data, while minimizing the false alarms that can become a distraction from the real opportunities for improvement. Statistical methods are important throughout the lifecycle supporting design space experimentation, PPQ sample size justification, and monitoring throughout the commercial life in CPV.

ADDITIONAL SUMMARY NOTES FROM TIM and JULIA (PROVIDED AFTER THE FORUM) Regarding the 3-lot discussion, a comment was made that it is useful to exclude failures. However it was highlighted that this isn t evidence that the process is capable. In addition 30-lots may not be the secret sauce, particularly if you have campaign effects or special cause variation which hasn t yet been experienced. We touched on the concept of replacing the Stage 2 verification plan with a strategic Stage 3 CPV plan. While there wasn t common acceptance of this between industry and regulators there is value in exploring it further from the point of view of Lifecycle Management from late stage development through to commercial. PPQ could be a special case of showing comparability between development process and commercial process much like there will be comparability exercises throughout the commercial lifecycle. Likewise other models for validation such as used for bioanalytical methods might be considered in the framework of Stage 2 validation. Susan Kirshner (FDA) mentioned in the context of blending Stage 2 with Stage 3 to fulfill the spirit of validation. Since we don t typically vary the long term factors such as raw materials during Stage 2 validation, we don t capture this. So we might update our understanding of the long term variability during Stage 3 when these factors come into play. In bioanalytical method validation they have the situation where they don t have in study samples during the method validations, so they complement validation samples with true patient samples afterwards