implementation of qbd based control strategy during drug
TRANSCRIPT
Implementation of QbD based Control Strategy during Drug Product Continuous Manufacturing
Frantz Elbaz
DDF summit Berlin, March 12th, 2018
TRD
Business Use Only
Overall QbD strategy
Define Target Product Quality (QTPP with list of CQA)
Identify Knowledge Baseline
Develop Product/Process Understanding
Define Design Space (for CPPs and CMAs)
Control Strategy (for CPPs and CMAs)
Quality Monitoring
ImprovementContinual
The Matterhorn, CH
2
QTPP and CQAs
Practical definitions :– QTPP : list of key characteristics that could impact the DP in-vivo
performance
– CQA : a quantifiable property of the DP that is considered as critical for achieving the DP intended in vivo performance (purity, efficacy and safety).
For a solid oral dosage form :➢ Aspect
➢ Assay
➢ Content Uniformity
➢ Degradation/Stability
➢ Dissolution rate , ...
Only the CQAs that could be impacted by formulation and/or process variables are considered in the RA.
Business Use Only3
Product/Process understanding
Developed based on the outcome of the first RA– Identification of the pCPPs and pCMAs
– Use of screening DoE and standalone trials to confirm (or not) criticality of
identified pCPPs and pCMAs
– Use of factorial DoE to define the Design Space for selected CPPs and CMAs
where regulatory flexibility is targeted
– Elaboration of specific control strategies for identified CPPs and CMAs (could
be : double check in BR, process control systems, PAT tools, regulation
loops, ...etc)
Specifically for CM, this understanding could be
conducted at “full” scale (commercial equipment train)
and then avoid any scale-up.
Business Use Only4
Continuous Manufacturing DP processing at Novartis
Business Use Only5
1. blending
A B
2. granulating 3. drying 4. sieving 5. tabletting 6. coating
Model compound formulated (25% DL) with common excipients
PAT instruments plugged at several locations to monitor and
maintain the process in a state of control (BU, LOD, PSD, CU) via
regulation loops to be implemented
First step was to focus our efforts on granules LOD after drying
Product/Process understanding
Example of the screening DoE that was conducted
Business Use Only6
7 Factors selected based on Fishbone analysis and Criticality Matrix of process
# Parameter to investigate Abbr. unit
1 solid feed rate SFR kg/h
2 liquid feed rate LFR kg/h
3 screw speed TSG SST rpm
4 barrel temp TSG BTT °C
5 dryer rotation speed FBD DRF rph
6 drying temperature FBD DTF °C
7 drying air volume FBD DAF m3/h
8* Room Temperature RT °C
9* drying inlet humidity FBD (=room humidity) DIH g/kg
10* pre-blend LOD PBL %
11* Product filter pressure at start of experiment (=influenced by runtime) PFS Pa
Screening-DoE:
Fractional Factorial
Design with 27-3 = 16
experiments and 3
center points
(Resolution IV)
And keep some pCPPs and pCMAs constant (screw design, DS characteristics,
Excipients characteristics....)
* "uncontrolled" DoE-parameter will be monitored during trials
Example : LOD of «dry» granules
Business Use Only7
Why focus was on LOD? Because granules LOD after drying impacts several CQAs
One full rotation sampled, each sample measured 3
times (error bars = variation within sample)
Starting point : fixed parameters used
for wet granulation and drying modules
Conclusions :
➢ LOD (measured by sampling and halogen
moisture analzer) equilibrated after 120
minutes
➢ LOD in acceptance range (LOD of blend ±
0.5%) after 30 minutes
➢ Need to shorten both the « wasting » and
« equilibration » durations.
DoEs conclusions
Business Use Only8
LOD
Liquid Feed Rate
Dryer Rotation Speed
Solid Feed Rate
Drying Temp
Drying Air Volume
SFR*LFR
SDoE: identification and
quantification of main
effects (CPPs)
ODoE: process models
for control of responses
established (e.g. LOD)
LOD = 1.225+0.072*DRS
Model verification
Business Use Only9
Adapting manually the dryer rotation speed (DRS identified as CPP), based on DoE-based
statistical descriptive model
Model based control demonstrated, accuracy is limited by models confidence interval and the common
variability in the process
Overcompensation could be avoided by combining model-based control with direct PAT-feedback
DoE ressources consumption
Comparison between CM and batch processes :– 16 trials
– Wet granulation / drying process + sieving + tableting for CM
– Wet granulation / drying process + sieving + external phase addition + final
blending (lubrication) + tableting for batch mode
– 40% DL
– CM allows a significant decrease of DS and FTE needs.
Business Use Only10
DoE CM DoE batch mode
Batch size 2 hours run (6 kg of blend) Given by the wet
granulator bowl size, dryer
bowl size, container size :
16 kg of blend
Total duration 2 weeks 6 weeks
DS consumption 38.4 kg 102.4 kg
Conclusions and next steps
• Conclusion on one model compound
– Example of effective QbD implementation in CM
– Demonstration of feasibility and verification of the model for one responseimpacting at least one DP CQA
• Further application possibilities
➢ Using this model compound :
– Generate other offline and online models for other responses (BU, granules PSD, tablets characteristics including CU)
– Implement control actions in the automation system to ensure that the process is maintained in its state of control
➢Extend this approach to other compounds (NCE and marketed products)
Business Use Only11
Acknowledgements• Victoria Pauli
• Markus Krumme
• Peter Kleinebudde
• Simon Ensslin
• Florence Desvignes
• Maud Duchesne
• Samuel Fischer
• Ahmad Mohamad
• Julien Taillemite
• Nhat Quang Nguyen Trung
• Laurent Pellegatti
• Yves Roggo
• Philipp Heger
• Hervé Notter
Business Use Only12
Abbreviations used• BR : batch record
• BU : Blend Uniformity
• CM : Continuous Manufacturing
• CMA : Critical Material Attribute
• CPP : Critical Process Parameter
• CQA : Critical Quality Attribute
• CU : Content Uniformity
• DL : Drug Load
• DoE : Design of Experiments
• DP : Drug Product
• FBD : Fluid Bed Dryer
• LOD : Loss On Drying
• PAT : Process Analytical Technology
• pCMA : potential Critical Material Attribute
• pCPP : potential Critical Process Parameter
• PSD : Particle Size Distribution
• QTPP : Quality Target Product Profile
• RA : Risk Assessment
Business Use Only13