Process Analyzer Continuous Verification

Transcription

1 Process Analyzer Continuous Verification Jason E. Dickens, PhD Property of GlaxoSmithKline

2 2 PAT Cost-Complexity-Benefit Continuum Low End High End Simple Sensors Univariate Analyzers Spectral Analyzers Analyzer Systems ph Conductivity DO2/DCO2 Turbidly Handhelds Viscometers Photometers Gas Analyzers NIR IR Raman Imaging Etc. On-line HPLC On-line titrators On-line SSs Etc.

3 3 Process Instrument Continuous Verification QbD Design Intent Real-time control (RTC) Real-time release (RTR) Real-time assurance (RTA) PAT Method Types Quantitative Semi-Quantitative Qualitative Real-time monitoring (RTM) Exploratory Increasing need for Process analyzer Continuous verification (PACV)

4 Statistical Process Control (SPC) 101 Inputs Process Outputs Variable Materials (CpK>1) Control CPP (Cpk>1) Variable (CpK<1) Controlled Materials (CpK>1) Control CPP (Cpk>1 Variable (CpK~1) Controlled Materials (CpK>1) Control CQA (Cpk>1) Controlled (CpK>1)

5 PAT Application Elements and Relative Readiness Define Develop Process Understand Control & Release Explore RTM RTA RTC/RTR Analyzer Sample Interface Calibration Model Method Method QC/QA (CV) Data Historian Reporting System IQ/OQ Instrument Control Method Capabilities PQ Pilot Plant or Live Process Live Process 5

6 6 Lab Analytics vs. Process Analyzer Lab Instrument Automation Manual or Semi-Auto Auto Continuous No Yes Oversight Yes No Environment Controlled Variable Suitability checks Point of Use Periodic Process Analyzer Sampling Grab Sample In-line and On-line Sample Prep. Yes for on-line only Calibration Extrinsic Empirical Diagnostics Yes Yes but available & relevant? Quality Well Defined Depends Validation Conventional Conventional (Why?)

7 7 Conventional Laboratory Analytical Method QA/QC Instrument diagnostics Calibration Standard Blank Standard Reference standard(s) Semi-automated ±Spec Limit or SPC QA/QC sample analysis Comprehensive report Complete Turn Key Solution with high level of QA/QC

8 Process Instrument Vendor Supplied Process Instrument Reference Standards IQOQ Prediction Engine (3 rd Party) Expert mechanical Integration (3 rd Party) Inst Control & data Management (Depends) Meaningful analyzer quality system (Gap) PAT Team Key Actions Measurement system Analysis Practical considerations PAT Method & System Validation Integration quality and process effects PAT Method & Automation Specs. QC approaches QA approaches Continuous Verification Few Turn Key Solutions 8

9 9 PAT Quality Problem Statement PAT QC is a multivariate issue but is currently addressed by univariate means Analytical Specification limits Other than environmental and integration factors, PAT QA is often overlooked Details Lack of sufficient QC metrics Vendor QC protocols and intrinsic diagnostics alone are insufficient Annual or semi-annual qualification checks too infrequent Chemometric diagnsotics (DModX and T2) only identify there s a problem Not sufficient for holistic root cause Insufficient QA approach Adoption prospective quality (i.e., control charts) knowledge driven QA automation protocols Analyzer Availability (i.e., MTBF & MTTR) is often unknown or provided by the vendor

10 10 Process Analyzer Continuous Verification Road Map Measurement System Analysis (MSA) Risk Analysis CV Instrument diagnostics Routine suitability Model diagnostics QA controls Sufficient QC metrics QA approach Mitigation Strategy Robustness Availability (MTBF & MTTR) Actual failure modes Effective PAT continuous verification (CV) affords A preemptive action to resolve quality problems before they occur Critical data required for route cause analysis during QIs

11 11 Key Process Analyzer Quality Questions What are the critical sources of error and faults for a PAT application? Are the quality tools within process analyzer systems adequate? Are Chemometric diagnostics sufficient as a QC metric? What are the CV elements for a PAT application?

12 Measurement System Analysis (MSA) PAT MSA σ Total Error = σ + σ + σ + σ SE SP SI S σ 2 D

13 13 Systematic Errors PAT Approach At-line In-line On-line Systemic Errors Process extraction Sample preparation In situ probe effects Process effects Fouling Interface faults Sample extraction Sample preparation Process effects Internal/External Ref. Material residuals

14 14 Key Quality Process Analyzer Questions What are the critical sources of error and faults for a PAT application? Are the quality tools within process analyzer systems adequate? Are Chemometric diagnostics sufficient as a QC metric? What are the CV elements for a PAT application?

15 15 Standard Vendor QC Metrics Example: Spectroscopic based process analyzer Intrinsic Signal-to-Noise test Signal peak test Energy test Extrinsic (Required external Ref. Sts.) Reference or Background tests Wavelength accuracy test Photometric reproducibility test Sensitivity & spectroscopic detector linearity tests Resolution test Scan & alignment tests Analyzer metrics assess instrument quality and are only indirectly relevant to the PAT method s quality.

16 External References Standards INTERNAL QC Standard Analytical Performance Standards (APS) Internal prepared QC standard PAT irrelevant CERTIFIED Reference Standards Vendor Supplied (Current State) For QC and/or external standardization Application Specific

17 Issue to Certified Reference Standards Cert. Ref. Std. PAT Method LDR Certified reference standards may not be relevant to the linear dynamic range of the method s calibration.

18 18 Supporting PAT Architecture (e.g., SIPAT) Autonomous Data collection Data mining Real-time prediction of quality parameters Real-time model diagnostics Monitoring and control Supports easy and fast process validation 21CFR11 ready Online visualization as well as reporting and analysis of historical data Integration to both existing and new infrastructures (PCS, Batch Engine, MES, LIMS, data portals, KMS, etc. Sufficient PAT QC metrics and QA tools? Except Chemometric/multivariate quality diagnostics Spectral-score and residual-score space diagnostics

19 19 Key Process Analyzer Quality Questions What are the critical sources of error and faults for a PAT application? Are the quality tools within process analyzer systems adequate? Are Chemometric diagnostics sufficient as a QC metric? What are the necessary CV elements for a PAT application?

20 Chemometrics Diagnostics Spectroscopic process analyzers which are full spectrum and often depend upon chemometric and MVA diagnostics Current Result Score-Space Diagnostic (e.g., Hotelling T 2 ) Detects model extrapolation risk Residual-Space Diagnostic (e.g., DModX or Q) DLR Detects unexplained variation in the acquired spectrum Is the deviation significant to the measured quality attribute? In either case, is the detected deviation does not identify the root cause. i.e., Does the deviation related to the Mfc process or the process analyzer system? 20

21 21 Key Process Analyzer Quality Questions What are the critical sources of error and faults for a PAT application? Are the quality tools within process analyzer systems adequate? Are Chemometric diagnostics sufficient as a QC metric? What are the CV elements for a PAT application?

22 22 Process Analyzer QC Elements Analyzer Intrinsic diagnostics Analyzer extrinsic diagnostics MVA Calibration model diagnostics Current State System and Method validation Analytical performance standards (as required) Desired State Relevant Control Charts Analyzer Availability (MTBF and MTTR)

23 23 Process Analyzer Quality Assurance Elements Routine process analyzer maintenance plan Automated instrument controls Current State Instrinsic Analyzer and Chemo diagnostics alarming Autonomous & Semi- Autonomous QC protocols Desired State Prospective process analyzer quality monitoring Routine review of PAT method quality control charts Control chart auto-alarming (i.e., trends)

24 Process Analyzer Prospective Quality Step change SPC or MSPC capture a systems overall quality and provide prospective quality. Pass/Fail performance Limits do not! Bias Drift In control Daily QC data

25 3 key Elements of Process Analyzer Continuous Verification Level 1: General Performance Critical Inst. Diagnostics % Fat (CID) % fat /01/02 24/01/02 03/02/02 13/02/02 23/02/02 05/03/02 15/03/02 25/03/02 04/04/02 14/04/02 24/04/02 date Extrinsic Inst. Attributes (EIA) 3.30 % Fat Level 2: Model Diagnostics (e.g., T2 and DModX) 3.25 % fat Level 3: MSPC PAT Method Quality /01/02 24/01/02 03/02/02 13/02/02 23/02/02 05/03/02 15/03/02 25/03/02 04/04/02 14/04/02 24/04/02 date Analytical Standard (AS) Quality PAT Method OOS PAT Method? 25

26 Conclusion Process Analyzer CV underpins the confidence of a PAT application Validation is only a snap shot in time. Adoption of Analytical performance standards or reference method as required Adoption of prospective quality Adoption of MSPC based PAT method quality comprising of: Analyzer diagnostics External reference (standard or ref. method) Chemometric diagnostics Availability is a critical robustness metric I would not give a fig for the simplicity this side of complexity, but I would give my life for the simplicity on the other side of complexity. --Oliver Wendell Holmes, Jr. 26

27 Questions? 27

28 28 Adoption road blocks for PAT What is limiting the use of PAT? Fit for purpose process instruments Multivariate model development & maintenance Perception issue New, Unknown, complicated Insufficient data volume Black Box Mentality Limited Comprehensive Commercial Support Effective process instrument continuous verification (PICV)