DEMONSTRATING CAPABILITY TO COMPLY WITH …

1 DEMONSTRATING CAPABILITY TO COMPLY WITH A TEST PROCEDURE: THE content uniformity AND dissolution acceptance limits ( cudal ) approach Jim Bergum September 12, 20111 Key Responses For Batch Release Potency (Average Drug Substance/Dosage Unit) dissolution (% Drug Substance Released at Specified Time) content UniformityUSP <905> The degree of uniformity in the amount of the drug substance among dosage units. 2 Harmonized uniformity of Dosage Unit (UDU) TestAll measurements of dosage units and criteria values are in percentage label claim (%LC). At each stage calculate the sample average and the sample standard deviation testedPass stage if:S110|M -| + , where M is defined ) |M -| + using all 30 results (S1+ S2)ii) No dosage unit is outside the maximum allowed range of *M to *M.

2 M is defined as follows:(i)If is less than , then M = (ii)If is between and , then M = .(iii)If is greater than , then M = ZoneXXXXXX3 cudal Test: Background Methodology Developed in Mid 80 s Application: Process Validation (Show process does what it purports to do) Show Specific Quality Attributes will meet associated Testing Standards (eg: CU) content uniformity (units have similar amount of drug) dissolution (units dissolve at required rate) Request/Mission - Develop limits based on the process validation sample results that provide confidence that the testing standard samples will pass the testing Limit Table Example for UDU TestMean (%LC)RSD(%) Relative Standard Deviation (RSD) Limit assures, with 90% confidence, that a future testing standard sample take from the batch has greater than a 95% chance of passing the UDU Confidence Interval, 95% Coverage, n=305 acceptance Limit Sample Results.

3 Mean = (%) = LimitJustification Provides high assurance that batch meets regulatory standard Assurance increases with increased sample size. Will always need a standard to define Acceptable . Can be used for more than validation or product release (ex: evaluation of NIR methods for CU used in real time release). Tied directly to regulatory requirements Ensures compliance with 21 CFR (d) - Testing and Release for distribution Can be used as a tool to meet the expectations set forth by FDA's Process Validation Guidance6 Final Statistical Statement If acceptance limit sample passed limits , then with a pre-specified confidence level, the probability that a testing standard sample will pass the testing standard is greater than a pre-specified coverage value.

4 Example: With 90% confidence, a future testing standard sample take from the batch has greater than a 95% chance of passing the testing Part Testing Standard (Ex: UDU) probability distribution for individual observations (ex: Normal with parameters (Mu) & (Standard Deviation)) known distribution parameters, mathematically derive* the Lower Bound for each stage (Note: Each stage may have multiple criteria!) This is the hard part! bound for overall test is the maximum of the individual stage lower bounds*Bergum, and Hua Li, " acceptance limits for the New ICH USP 29 ContentUniformity Test," Pharmaceutical Technology, October 2007, pp. Test - 95% Lower Bound Contour9 Batch Standard DeviationBatch MeanProb(Passing UDU)MeanStd DevSimulate USP (N=1,000,000)90, , Sample/Criteria Types acceptance Limit Sample(Results Compared to acceptance Limit Table Criteria) Sampling Plan 1: One unit per location Sampling Plan 2: n units per location (allows estimation of between/within location variability) UDU Sample(Results Compared to Testing UDU Criteria) Sampling Plan: Defined by the Testing Standard (Usually a random sample from the batch Sampling Plan 1)10 Strategy Part Sampling Plan (1 or 2).

5 Confidence interval for the distribution parameters based on user defined confidence lower bound probabilities for each point in the confidence maximum probability across all points in confidence interval. maximum probability to user defined coverage (Lower Bound).11 UDU Test Lower Bound with Simultaneous Confidence IntervalBatch MeanULS = Upper CI for Batch SD95% Lower BoundBatch Standard Deviation12( , S) ( Z*ULS/ n, ULS)XConfidence IntervalXConstruct acceptance Limit Table To Generate Table, user selects Confidence Level (Usually 90 or 95%) Coverage Lower Bound - Desired Probability of future Testing Standard Samples passing Testing Standard (usually 95%).

6 Sampling Plan/Sample Size Possible values specific to the Testing Standard (ex: Q for dissolution , Target for content uniformity ) 13 acceptance Limit Table(Sampling Plan 1, 90% CI, 95% Coverage, n=30)MeanRSD(%) Size Effect on RSD LimitSampling Plan 1 SampleMean(%LC)RSD(%) acceptance Limit Table- Determining Adequate Sample Size Pick sample size such that the probability of passing the acceptance limit table is: High for a Good Batch (Acceptable Batch Mean and Std Dev) Low for a Bad Batch (Unacceptable Batch Mean and Std Dev) For given sample size, find probability of passing acceptance limit table by integrating over tabled : Sampling Plan 1P(Passing acceptance Limit Table)Batch MeanBatch RSDS ample Plan 1 Operating Characteristic Curve (OC)USPL arge NCuDAL18 Batch Mean = 100%LCUSPCuDAL N = 10 cudal N= 100 cudal N= 300 Sampling Plan 1 Operating Characteristic (OC) Curve19 Batch Mean = 96%LCUSPCuDAL N = 10 cudal N= 100 cudal N= 300 Example: Sampling Plan 2 (15 X 4)20 ResultSummary statisticsLocation1234 MeanVarianceStd : Sampling Plan 2 21 Example.

7 Sampling Plan 222 Descriptive (within-location StdDev) deviation of location Deviation of Location ULLL ULLL ULLL Plan 2:Sample Size Evaluation23 Prob(PassingAcceptance Limit Table)Batch MeanVariance Component (SD)Sampling Plan (Loc x #/Loc) Between Location*(Example= )Within Location(Example= ) * = ((SD Location Means)2- SE2/n)Current Status General Methodology: ASTM E11 Standard E2709 Original (Done) Lower Bound Confidence Intervals acceptance Limit Table Revision (In 2ndBallot) Add Sampling Plan 2 content uniformity : ASTM E55 Standard (3rdBallot) Lower Bound Calculations Specific to UDU Test Future Other Tests?

8 24