This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
FOR SOLID ORAL DOSAGE FORM DEVELOPMENT AS PER QbD
OPTIMIZATION OF CRITICAL PROCESSING PARAMETERS OF DRY MIXING PROCESS
OBJECTIVE of the experiment & NUMBERS of the factors involved are the primary two most important factors required to be considered during selection of any design for experimentation.
CA
SE
STU
DY
“High”
“Medium”
“Low”
• In Dry Mixing Process, 2 Processing Parameters were critical & required to be optimized
• Moreover, It was required to investigate interactive & quadratic relationship between factors & response to find out optimum ranges
• Thus, 3 Level FFD is a time & cost effective best option for optimization of 2 factors.
• However 3 Level FFD facilitates investigation of interactive & quadratic relationship of factors & response in the terms of multiplied 2FI & squared main effects in the quadratic model equation
CPPs CQAs
PLACKETTE BURMAN
2 LEVEL FACTORIAL
3 LEVEL FULL FACTORIAL
OPTIMIZATION OF CRITICAL PROCESSING VARIABLES OF DRY MIXING/ BLENDING PROCESS
During Selection of order of polynomial: MODEL (A mathematical relationship between factors & response assisting in calculations & predictions) for Analysis of Response; ANOVA was carried out thoroughly for testing of SIGNIFICANCE of every possible MODEL (p<0.05), insignificant LACK OF FIT (p>0.1)
with response surface to confirm expected shape of response behavior
P-Value < 0.05 (Significant) P-Value > 0.10 (Insignificant) Lack of Fit is the variation of the data around the fitted model. If the model does not fit the actual response behavior well, this will be significant. Thus those models could not be used as a predictor of the response.
P-Value < 0.05 (Significant) P-Value > 0.10 (Insignificant) Sequential model sum of square provides a sequential comparison of models showing the statistical significance of
ADDING new model terms to those terms already in the model. Thus, the highest degree quadratic model was selected having p-value (Prob > F) that is lower than chosen level of significance (p = 0.05)
Sequential MODEL Sum of Square Tables
LACK of Fit Tests
Response 1: Average Assay in BU Response 2: %RSD in BU
CA
SE
STU
DY
Response 1: Average Assay in BU Response 2: %RSD in BU
PREDICTION EFFECT EQUATION ON INDIVIDUAL RESPONSE BY QUADRATIC MODEL
Average Assay of Blend Uniformity =+99.61 +0.78A+2.32B-0.95AB-1.52A2-2.22B2
RSD Of Blend Uniformity=+1.94-0.47A-1.45B+0.53AB+1.13A2+1.98B2
PLACKETTE BURMAN
2 LEVEL FACTORIAL
3 LEVEL FULL FACTORIAL
OPTIMIZATION OF CRITICAL PROCESSING VARIABLES OF DRY MIXING/ BLENDING PROCESS
Numerical Analysis of Model Variance was carried out to confirm or validate that the MODEL ASSUMPTIONS for the response behavior were met with actual response behavior or not, via testing of significance of each MODEL TERMs
with F >>1 & p<0.05, insignificant LACK OF FIT (p>0.10), adequate PRECISION > 4, R2 Adj & R2 Pred in good agreement <0.2d as NUMERICAL INDICATORS, with well behaved RESIDUALS analyzed by diagnostic plots as GRAPHICAL INDICATORS.
Residual (Experimental Error) Noise = (Observed Responses) Actual Data– (Predicted Responses) Model Value During RESIDUAL ANALYSIS, model predicted values were found higher than actual & lower than actual with equal probability in
Actual Vs Predicted Plot. In addition the level of error were independent of when the observation occurred in RESIDUALS Vs RUN PLOT, the size of the observation being predicted in Residuals Vs Predicted Plot or
even the factor setting involved in making the prediction in Residual Vs Factor Plot
CA
SE
STU
DY
Response 1: Average Assay in BU Response 2: %RSD in BU
PLACKETTE BURMAN
2 LEVEL FACTORIAL
3 LEVEL FULL FACTORIAL
OPTIMIZATION OF CRITICAL PROCESSING VARIABLES OF DRY MIXING/ BLENDING PROCESS
Response 1: Average Assay in BU Response 2: %RSD in BU
Model Graphs gave a clear picture of how the response will behave at different levels of factors at a time in 2D & 3D
Factors (Variables) Knowledge Space Design Space Control Space A Blending Speed (RPM) 8.0-12.0 9.15-11.35 9.5-11.0 B Blending Time (minutes) 5.0-15.0 10.0-13.5 10.0-12.0
Responses (Effects) Goals for Individual Responses Y1 Avg. Assay of BU (%) To achieve average assay of BU in the range from 98 to 102%
Y2 RSD of BU(%) To achieve minimum variability in BU i.e. NMT2.0%
PLACKETTE BURMAN
2 LEVEL FACTORIAL
3 LEVEL FULL FACTORIAL
OPTIMIZATION OF CRITICAL PROCESSING VARIABLES OF DRY MIXING/ BLENDING PROCESS
By Overlaying contour maps from each responses on top of each other, RSM was used to find out the IDEAL “WINDOW” of operability-Design Space per proven acceptable ranges & Edges of Failure with respect to individual goals
CA
SE
STU
DY
PLACKETTE BURMAN
2 LEVEL FACTORIAL
3 LEVEL FULL FACTORIAL
OPTIMIZATION OF CRITICAL PROCESSING VARIABLES OF DRY MIXING/ BLENDING PROCESS
After completion of all experiments according to DoE, Verification was required TO CONFIRM DESIGN SPACE developed by selected DESIGN MODEL , which should be rugged & robust to normal variation within a SWEET SPOT in OVERLAY PLOT,
where all the specifications for the individual responses (CQAs) met to the predefined targets (QTPP)
8.0-12.0
9.15-11.35
9.5-11.0
5.0-15.0
10.0-13.5
10.0-12.0
The OBSERVED EXPERIMENTAL RESULTS of 3 additional confirmatory runs across the entire design space were compared with PREDICTED RESULTS from Model equation by CORRELATION COEFFICIENTs. In the case of all
3 responses, R2 were found to be more than 0.900, confirming right selection of DESIGN MODEL.
BLENDING SPEED (RPM) BLENDING TIME (MIN)
KNOWLEDEGE SPACE
DESIGN SPACE
CONTROL SPACE
Known Ranges of OPERABILITY before Designing
Optimized Ranges of FEASIBILITY after Development
Planned Ranges of CONTROLLING during Commercialization
Quality Risk Manager & Intellectual Property Sentinel- CIIE, IIM Ahmedabad MS (Pharmaceutics)- National Institute of Pharmaceutical Education & Research (NIPER), INDIA
PGD (Patents Law)- National academy of Legal Studies & Research (NALSAR), INDIA