Practical Considerations for Implementation of Real depts. ?· Practical Considerations for Implementation…

Download Practical Considerations for Implementation of Real depts. ?· Practical Considerations for Implementation…

Post on 04-Sep-2018




0 download

Embed Size (px)


<ul><li><p>Practical Considerations for Implementation of Real Time Release testing (RTRt)</p><p>Graham D. Cook Ph.D.,Senior Director, Global Quality OperationsHavant,</p><p>CPAC Satellite MeetingRome, March 25, 2010</p></li><li><p>1</p><p>Overview</p><p>What needs to be considered when implementing Real Time Release Testing?</p><p>Case Study 1 Product X</p><p>Case Study 2 Product Y</p><p>Comparison of some aspects of Product X and Y</p><p>Conclusions</p></li><li><p>2</p><p>Real Time Release Testing</p><p>The ability to evaluate and ensure the quality of in-process and/or final product based on process data, which typically include a valid combination of measured material attributes and process controls. (ICH Q8(R2))</p></li><li><p>33</p><p>REAL TIME RELEASE TESTING</p><p>SystemsScience</p><p>People</p><p>ICH Q10 Pharmaceutical Quality System</p><p>ICH Q8 Pharmaceutical Development</p><p>ICH Q9 Quality Risk Management</p><p>Real Time Release Testing - Essential Elements</p></li><li><p>44</p><p>ScienceProcess data and Control strategy?Analytical methods and specifications?</p><p>Sampling and StatisticsAcceptance criteria</p><p>PeopleOrganization and Training?</p><p>Pharmaceutical Quality SystemQuality risk management?Disaster recovery?Model maintenance?Handling of outliers?Data management? Batch disposition?</p><p>Regulatory Interactions</p><p>Considerations for Implementing Real Time Release Testing</p><p>Topic discussed in Case StudyProduct X Product Y</p><p>Note: All above factors considered for Product X and Product Y</p></li><li><p>5</p><p>Overview</p><p>What needs to be considered when implementing Real Time Release Testing?</p><p>Case Study 1 Product X</p><p>Case Study 2 Product Y</p><p>Comparison of some aspects of Product X and Y</p><p>Conclusions</p></li><li><p>666</p><p>Product X - Product Description</p><p>Dosage FormBCS Class 3 compound (High Water Solubility, Moderate Permeability)Monolithic Extended Release Tablets (multiple strengths)Relatively high dose compound (high drug load)Level A IVIVC has been established for all dose strengths</p><p>Polymer is the primary driver of clinical performanceVariation in process parameters did not impact the dissolution performance</p><p>Potential to apply Real Time Release TestingProcess understanding at pilot-scale translated well to commercial manufacturing scaleRobust control strategy </p></li><li><p>7</p><p>Product X - Manufacturing Process</p><p>Robust control strategy Increased assurance of quality RTRt</p><p>PAT1 PAT</p><p>1</p><p>PAT:1 = Blend Uniformity2 = Granule particle size3 = Weight, Hardness, Potency, Drug concentration, Identity, </p><p>Ratecontrolling polymer concentration</p><p>PAT2 PAT</p><p>3</p><p>Holistic Control Strategy e.g.:Content Uniformity = Blend uniformity + Drug concentration</p><p>+ Weight control</p></li><li><p>88</p><p>8</p><p>Development of Sampling Plans</p><p>PAT AnalyzersMeasure condition of the process material in real timeCollect more information about the batch</p><p>When and where do we sample?Use of prior knowledge of product and risk assessment in design of plan</p><p>e.g. consider potential for segregation (failure mode) in a compression mixStatistical rationale for sampling plan</p><p>Performance of plan relative to assurance of qualityRelationship between Real Time Release Testing and Pharmacopoeial TestingPlacement of PAT device in manufacturing equipment</p><p>Ensure analyzer measurements are representative of the batch</p></li><li><p>9</p><p>Operating Characteristic CurvesPlot the probability of accepting a batch against the quality level</p><p>What is the probability of rejecting a good batch? (Producer risk)What is the probability of releasing a bad batch? (Consumer risk)</p><p>Operating Characteristic curves enableDifferent batch acceptance sampling plans to be investigatedDesign of a batch sampling plan with the desired performance</p><p>N = batch size</p><p>n = sample size </p><p>c = critical acceptance number</p><p>AQL = Acceptable Quality Level</p><p>LQL = Limiting Quality Level</p><p>Steepness of the curve indicates discriminating ability of the plan</p></li><li><p>1010</p><p>Operating Characteristic Curves for UDU Test</p><p>The USP OC curve shows 90% probability of passing a batch with 2% product outside 85-115% of Label ClaimThe Alternative Plan OC curve shows 0% probability of passing a batch with 2% outside 85-115% of Label Claim The Alternative Plan uses a larger sample size and provides a better assurance of quality than USP</p><p>ProbabilitytopassAlternativePlan90%probabilityofpassingbxbyUSPProbabilitytopassUSP</p><p>%Productoutside85115%ofLabelClaim</p><p>0 1 2 3 4 5 6 7</p><p>Note: UDU = Uniformity of Dosage Units (Content Uniformity)Coverage = % Product meeting 85-115% of Label Claim Coverage term used in paper by Sandell, D. et al. (see reference slide 24 )</p><p>If the Alternative Plan is chosen for sampling using a PAT analyzer what is the fall-back if the analyzer fails?</p><p>- Must maintain same level of quality by using the same sampling plan with the replacement analytical technology</p></li><li><p>11</p><p>Product X - Sampling</p><p>Location of PAT Analyzer in BlenderRationale based on prior knowledge (including publications)Blender design: no hot spots therefore any location suitableAnalyser placed in bottom of blender</p><p>Sampling plan for TabletsTablets samples taken uniformly across batchX tablets taken every Y minutes during compression run</p><p>Drug Content of TabletsDrug concentration by NIR and Weight Uniformity determined on different tablet samples</p><p>Low risk because high drug loadingCheck using Monte Carlo simulation</p></li><li><p>12</p><p>Monte Carlo SimulationsMonte Carlo Simulation* </p><p>A technique that converts uncertainties in input variables of a model into probability distributions. By combining the distributions and randomly selecting values from them, it recalculates the simulated model many times and brings out the probability of the outputAllows several inputs to be used at the same time to create the probability distribution of one or more outputsThe output is generated as a range instead of a fixed value and shows how likely the output value is to occur in the range</p><p>Assessment of Drug Content of TabletsDrug Concentration (NIR) and Tablet Weights are independent probability distributionsWhat is probability of releasing a batch of marginal quality?Use Monte Carlo simulation to explore the variation in:</p><p>Drug Concentration (NIR mean; NIR Weight (Weight mean; Weight std. dev.; difference in weights from double-sided tablet press)</p><p>*Sanford Bolton, Charles Bon: Pharmaceutical Statistics - Practical and Clinical Applications, Fourth Ed., Marcel Dekker</p></li><li><p>13</p><p>Using Monte Carlo Simulations to Assess RiskAssessment of risk of releasing a marginal quality batch for chosen Coverage level:Combination of Drug Concentration (NIR mean) and Tablet Weight (WT mean) parameters</p><p>Indicates potential scenarios where there is a high probability of releasing a batch of marginal quality:cases with probability &lt; 6%cases with probability 6 - 8%cases with probability 8 - 10%cases with probability &gt; 10%</p><p>+</p><p>Monte Carlo simulations: 1400 cases, 5000 simulations run Worst-case scenarios run</p><p>Very low number of high probability cases found equates to very low probability of releasing marginal quality batch</p><p>Simulations help provide an assessment of the risk for the chosen coverage level and sampling plan</p></li><li><p>1414</p><p>RTRt - OrganizationMulti-disciplinary / cross-functional teams are key to RTRt </p><p>New skill sets may be needed</p><p>Real Time Release testing (RTRt)</p><p>Technology</p><p>Regulatory Affairs</p><p>QualityOperations</p><p>Statistics</p><p>FormulationDevelopment</p><p>Chemometrics</p><p>AnalyticalDevelopment/PAT</p><p>Operations</p></li><li><p>1515</p><p>15</p><p>Impact on Quality Systems</p><p>Some considerationsWhat happens if parameters are outside -</p><p>Normal Operating Ranges (NORs)? Design space?</p><p>What do we do if a PAT measurement system stops functioning?How do we handle atypical results/outliers?</p><p>Cannot test into compliance using alternative methods (ICH IWG Q&amp;A)</p><p>What do we do if the chemometric model is no longer appropriate?What is the impact on the batch release process? </p><p>What are alternative sampling plans and measurement systems to enable disposition decisions?</p></li><li><p>16</p><p>Quality Risk Management</p><p>Quality Risk ManagementEnabler for implementation of Real Time Release testing </p><p>Incorporate QRM into proceduresFacilitates uniform implementation of QRM across the entire organizationConsistent use of the same language (terminology) and processEstablish criteria for re-evaluation of risks and mitigation plans</p><p>Triggered by time, event or new knowledge</p><p>Training program Different levels</p><p>Awareness trainingUser training - participant, facilitator, team leader</p><p>Facilitates selection of correct QRA approach and toolCulture - proactive approach to quality</p></li><li><p>1717</p><p>Decision Tree for Equipment Failure</p><p>* Note that it may be impractical to stop some unit operations (e.g. wet granulation) during processing </p><p>Does PAT pass System Suitability?</p><p>NO</p><p>Proceed with unit operation/manufacturing </p><p>process</p><p>YES</p><p>YES</p><p>NO</p><p>Is PAT functional during </p><p>the Process?</p><p>YES</p><p>Can the instrument be repaired in a suitable time frame? </p><p>Can the instrument be replaced with a spare instrument?</p><p>Generate Event Report Form, Fix/replace</p><p>instrument</p><p>NO</p><p>Are there alternative controls to ensure/control process variability?</p><p>Are there measurements downstreamwhich could be used to correlate data?</p><p>Do we need further sampling?</p><p>Generate Event Report Form, Capture Process/</p><p>Action Items</p><p>Revert to Testing using Regulatory Analytical Procedure </p><p>Generate Investigation/ERF to identify root cause</p><p>Capture Process/Action Items</p><p>YES</p><p>Stop Process*:Evaluate Instrument</p><p>Define response in advance = proactive quality, rather than after problem has occurred = reactive quality</p><p>Extract of Decision Tree for PAT equipment failure</p></li><li><p>18</p><p>Quality Systems and Processes-Handling of Outliers</p><p>Atypical Results/OutliersBad data or valuable information?Statistical processes subject to random variability</p><p>Pre-defined mechanisms or systems should be developed = proactive approach to quality</p><p>Process to define outliers or invalid dataDefinition of deviations that require investigationSpecifications must be carefully defined</p><p>Consideration of Impact on Product QualityUse science- and risk-based approaches Use an holistic assessment of all product and process measurements to assess qualityFrequency of outliers may be importantExtent of process knowledge and process/product history (e.g. changes)Potential impact to patient safety and efficacy</p></li><li><p>19</p><p>Product X Regulatory Interactions</p><p>FDAFDA CMC Pilot </p><p>February 2008 NDA ApprovedComparability Protocol for RTRt approved as part of the original application</p><p>July 2007 Pre-Operational Visit to manufacturing facilityOctober 2009 Prior Approval Supplement for RTRt submittedNovember 2009 PAI for RTRt</p><p>Pre-Operational VisitFDA personnel</p><p>Center (Reviewers (2), Compliance officer (1))Field (District (2), Patriot team (1))</p><p>Focus on PAT systems and Quality System includingProcedures for chemometric model maintenance Chemometric modelsEquipment failure recovery systemsSampling Plans Risk Assessments</p><p>(PAI similar to POV)</p></li><li><p>20</p><p>Overview</p><p>What needs to be considered when implementing Real Time Release Testing?</p><p>Case Study 1 Product X</p><p>Case Study 2 Product Y</p><p>Comparison of some aspects of Product X and Y</p><p>Conclusions</p></li><li><p>212121</p><p>Product Y - Product Background</p><p>Dosage FormBCS Class 1 compound (High Water Solubility, High Permeability)Immediate Release TabletsPotent, low dose compound, low drug load</p><p>Launched from small-scale containment manufacturing facility</p><p>Potential for Real Time Release TestingIncreased demand need to increase efficiency</p><p>More information in real-timeRobust control strategy</p></li><li><p>22</p><p>API &amp; Excipients</p><p>Dispensing Blend BlendSieving Granulation</p><p>MagStearate</p><p>MagStearate</p><p>PAT1</p><p>PAT2</p><p>BlendTablettingCoating PAT3</p><p>PAT4</p><p>PAT2</p><p>PAT2</p><p>PAT:1 = Identification2 = Blend Uniformity3 = Weight, Hardness, Disintegration, Potency, Content Uniformity 4 = Water Determination</p><p>Product Y - Manufacturing Process</p></li><li><p>23</p><p>Control StrategyMethods for conventional QC Release</p><p>TestingMethods for Real Time Release</p><p>Testing</p><p>Appearance (Visual) Appearance (Visual) </p><p>Identity (TLC / HPLC) Identity (NIR)</p><p>Impurities (HPLC) Not Tested routinely - due to process capability and understanding</p><p>Assay / CU (NIR + Weight)</p><p>Dissolution (Dissolution Tester) Disintegration (Disintegration Tester)</p><p>Water Determination (Karl Fischer) Water Determination (NIR)</p><p>Microbial Quality (Microbiology) Not Tested routinely - due to process capability and understanding</p><p>Assay / CU (HPLC)</p><p>Note: Real Time Release Testing does not mean less testing!</p></li><li><p>24</p><p>Specifications for Large n UDU testing</p><p>Specifications based on zero tolerance* problematic when testing large samples (i.e. &gt;30)Large n Counting Test**</p><p>Analyze X samplesAssess number of samples (n) outside of limitsCompare n to defined criteria (c) for allowed number of samples outside of limitsBatch is in control when n &lt; c</p><p>*Zero tolerance specifications (e.g. no tablet outside of 75 125% of target) discourage intensive sampling to gather process information because the larger the sample size, the greater the chance of an OOS result</p><p>**Sandell, D., Vukovinsky, K., Diener, M., Hofer, J., Pazdan, J. and Timmermans, J (2006), Development of a content uniformity test suitable for large sample sizes, Drug Information Journal 40 (3), 337 - 344.</p><p>100%75% 125%</p><p>Failing Tablets</p></li><li><p>25</p><p>Product Y - Sampling</p><p>Sampling plan for TabletsOC curves used to assess performance of sampling plansLarge n Counting Test adoptedTablets samples taken using stratified sampling plan across batchTablets samples taken at minimum of P points during compression runMinimum of Q tablet samples taken</p><p>Drug Content of TabletsTablet is assayedNIR for drug concentration and weight measured on the same tablet</p></li><li><p>26</p><p>31302928272625242322212019181716151413121110987654321</p><p>6,5</p><p>6,0</p><p>5,5</p><p>5,0</p><p>4,5</p><p>4,0</p><p>3,5</p><p>Location Number</p><p>CU S</p><p>ingl</p><p>e V</p><p>alue</p><p>s</p><p>3,75</p><p>6,25</p><p>4,25</p><p>5,75</p><p>Boxplot of CU in batch 810467400XXXXXXXXX</p><p>Product Y Sampling for Content Uniformity</p><p>125%</p><p>115%</p><p>473523118864c</p><p>1000750500250200183150100n</p><p>Large n Counting Test Criteria </p><p>n = number of tablets sampledc = acceptable number of tablets outside 85% to 115%</p><p>Large n counting test:</p><p>-- accounts for possibility of tablets outside 85% to 115% when collecting large samples</p></li><li><p>27</p><p>Chemometric Model Maintenance and Update</p><p>Periodic Evaluation of Applicability of NIR Model used for UDUNIR model evaluated with every batch (system suitability)Annual evaluation of model</p><p>Updating the ModelMechanism - Internal change control procedures used to identify material or process changes with potential to impact the NIR modelNIR measurements may not be valid</p><p>Operation outside NIR model limitsNew variability in materials and/or process Fall-back: reference HPLC method used for batch release</p><p>Assessment for model update:No impact:</p><p> New variability is unrelated to active (i.e. not significant in the regression coefficients) or is compensated for by model parameters. </p><p> Update model (without changes) to include robustness to this new variability Impact:</p><p> Update the model e.g. include new variability within the model or optimize calibration model parameters</p></li><li><p>28</p><p>RTRt - Training</p><p>Quality by Design TrainingGeneral conceptsSite-wide training Culture change?</p><p>Quality SystemQuality System Elements revised</p><p>DeviationsChange ManagementValidationKnowledge Management</p><p>Training on revised SOPs</p></li><li><p>29</p><p>Control Charts</p><p>Integrating PAT into Facility Data Management </p><p>10X Unit</p><p>NIR</p><p>Report</p><p>MES-Systems</p></li><li><p>30</p><p>Batch Disposition</p><p>Batch dispositionB...</p></li></ul>


View more >