Date post: | 22-Dec-2015 |
Category: |
Documents |
Upload: | hannah-hunt |
View: | 217 times |
Download: | 0 times |
Steven Novick, Katherine Giacoletti, Tara Scherder, and Bruno [email protected]
How to implement Bayesian statistics to Make Lifecycle Strategy a Reality that Serves Quality
Quality by Design overview
• Quality Target Product Profile (QTPP)
• Determine critical quality attributes (CQAs) and Specifications
• Perform risk assessment
• Develop a design space
• Design and implement a control strategy: SPC
• Manage product lifecycle, including continual improvement: Transfer
Prove the objectives will be met surely in the future
CQA’s
Product Profile
Risk Assessments
Design Space
Control Strategy
Continual Improvement
Stage1 Process Design
Stage 2Process
Performance Qualification
Stage 3Continued Process
Verification
the commercial process is defined based on knowledge
gained through development and scale-up activities
the process design is evaluated and assessed
to determine if the process is capable of
reproducible commercial manufacturing
ongoing assurance is gained during routine production that the
process remains in a state of control
SCIENTIFIC EVIDENCE ACROSS LIFECYCLE
2011 FDA Guidance Process Validation
Excerpt from Guidance
• “…high degree of assurance on the performance of the manufacturing process that will consistently produce….”
• “ ….collection and evaluation of data … which establishes scientific evidence that a process is capable of consistently delivering quality product….”
• “ … the assurance should be obtained from objective information and data from laboratory, pilot batches….”
Excerpt from Guidance
• “During the process qualification (PQ) stage of process validation, the process design is evaluated to determine if it is capable of reproducible commercial manufacture…”
What is “capable”?Capability is defined as the ability of a process to meet specification
Bayesian principle
TotalData
AvailableData
ObservedData =+
“LIKELIHOOD”data coming from the
experiment
“POSTERIOR DISTRIBUTION” for parameterscombination of information collected before the experiment
and what comes from the experimental data
“PRIOR DISTRIBUTION” from previous studies, expert
opinion, literature,…
• Uncertainty is described in terms of probability :
Bayesian principle
P(θ>5.5)=0.401
Bayesian principle
Bayes directly tests hypotheses: P(performance|data)
Frequentist method is indirect P(data|performance)
PRIOR distribution BATCH data POSTERIOR distribution
P(potency in Specs)= P(quality)
+
Bayesian principle
• PPQ batches are produced to collect evidence of the quality of the process
- Frequentist analysis:• Point estimate and confidence intervals as summaries of process
(mean and sd) What do PPQ batches tell us about the process?
- Bayesian analysis:• Before the PPQ: a priori opinion on the process How should those PPQ batches change our opinion about the
process? How should those PPQ batches provide assurance about future
batches?
• Motivations for adopting Bayesian approach: Natural and coherent way of thinking about learning and risk
How to make predictions
Monte-Carlo Simulations• “new observations” ỹ ~ F( ,m s)• ( ,m s) are [erroneously] fixed
Bayesian Predictions• “new observations” ỹ ~ F( ,m s)• ( ,m s) ~ (p m0,s0| data)
Bayesian Predictive Distribution
The Bayesian theory provides a definition of the Predictive Distribution of a new observation given past data.
Joint posteriorModelIntegrate over parameter distribution
MarginalModel Conditional
𝑝 (~𝑦|𝑑𝑎𝑡𝑎 )=∬𝜇 ,𝜎 2
❑
𝑝 (~𝑦∨𝜇 ,𝜎2 ,𝑑𝑎𝑡𝑎 )×𝑝 (𝜇 ,𝜎 2∨𝑑𝑎𝑡𝑎 )𝑑𝜇𝑑𝜎2
¿∬𝜇 ,𝜎 2
❑
𝑝 (~𝑦∨𝜇 ,𝜎 2 ,𝑑𝑎𝑡𝑎 )×𝑝 (𝜎 2∨𝑑𝑎𝑡𝑎)×𝑝 (𝜇∨𝜎2 ,𝑑𝑎𝑡𝑎 )𝑑𝜇𝑑𝜎 2
Integral typically computed by Monte
Carlo methods
Comparison Frequentist vs Bayesian
• When NON-informative priors are envisioned
- Posteriors and HPD (~quantiles) are the same as the Frequentist results
Are non-informative priors defensible in Stage 2?
There are defensible priors
• Once decision is made to go through PPQ, there is belief it will work.• Translate those scientific evidence and data based into priors
• Priors contain the whole uncertainty about this belief. This is the prior elicitation process.
• Classical statistics ignores prior available information.
Stage 2 and Bayesian Method
-∞ +∞
P
X X X X
X
X X X
Based on a point estimate of µ, σ Based on a distribution of µ and σ
PredictiveDistribution
Prior Distribution
PPQ batches
Frequentist Bayesian
Stage 2 and Bayesian Method
-∞ +∞
P
X X X X
X
X X X
Based on a point estimate of µ , σ Based on a distribution of µ and σ
PredictiveDistribution
Prior Distribution
PPQ batches
Frequentist Bayesian
Probability being in specificationsvs Tolerance intervals
• Use the Predictive distribution to compute the prob. to be within specs
X X X X
[---------------------] Tolerance Interval
Bayesian method directly calculates risk
Frequentist tolerance interval indirectly assesses risk
Predictive posterior
Number of Batches
• Number of batches required to guarantee 95% of success in PPQ, i.e. that 96% of future results will be within specifications.
Class
ical
Sta
tistic
s
Bayesian Statistics• Classical Stats requires more
than 10 batches• Bayesian statistics using prior
(defensible) information only requires 4 batches.
Why?
• The Posterior of performance parameters is more precise.
Other Benefits of Bayesian Approach
• Capability is defined as the ability of a process to meet specification, that is, the probability of meeting specification
Bayesian provides a true prediction of future performance
• Handles complicated hierarchy/ sampling plan- Between batch, sample within batch, within sample variation can
be incorporated- Unbalanced sampling
• Joint prediction of multiple CQAs is possible• Uncertainty of parameters included, thus improving prediction and
reducing risk• Not affected by non-centering within specification range• Systems approach to unit operations (simultaneous prediction)
Stage 1 - Design Space and Predictions
• In Stage 1 the objective is to identify the Design Space• DoE are performed to understand the relationships
between the CPP and the CQA
• The known or assumed control/uncertainty on CPPs can be integrated into Predictions
• The set of CPP (X) that guarantee results are in specifications is called the Design Space.
𝑝 (~𝑦|𝑑𝑎𝑡𝑎¿= ∫𝜇 ,𝜎 2
❑
∫𝑋
❑
𝑝 (~𝑦∨𝜇 ,𝜎2 ,𝑋 ,𝑑𝑎𝑡𝑎 )×𝑝 ( 𝑋 )×𝑝 (𝜇 ,𝜎2∨𝑑𝑎𝑡𝑎 )𝑑𝑋 𝑑𝜇𝑑𝜎2
An example: Spray-drying process
• Spray-drying is intended to create a powder with small and controlled particle size for pulmonary delivery of a drug substance
• Several Critical Process Parameters (CPP) have an influence on several Critical Quality Attributes (CQA)
- Inlet temperature- Spray flow-rate- Feed rate
(other process parameters are kept constant)
• Specifications on CQA defined as minimal satisfactory quality
- Yield > 80% - Moisture < 1%- Inhalable fraction > 60%- …
The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty by Dr. Sam Savage
Focusing only on the mean (average) can put us at risk!
Average depth of river is 3 feet.
From John Peterson, 2012
Spray-drying process
Risk-based design space: predicted P(CQAs ∈ l)•
In the Design Space, there is 45% of chance to observe each CQA within specification, jointly
~ 45% probability to jointly observe CQAs within specification
100-45% = 55% of risk not to observe the CQAs within specification (jointly) !
Inlet.Temperature
Spr
ay.F
low
.Rat
e
Inlet.Temperature
Fee
d.R
ate
Inlet.Temperature
Fee
d.R
ate
Spray-drying process
• Validation- Experiments have been repeated 3 times independently at optimal
condition, i.e.Inlet Temperature: 123.75°CSpray Flow Rate: 1744 L/hFeed Rate: 4.69 ml/min
Jointly, 2 out of the 3 runs within specification
Spray-drying process
• Post-analysis (« How they are statistically distributed »)- Marginal predictive densities of the CQAs
Inhalable fraction is predicted to be widely distributed
Predictive uncertainty = data uncertainty + model uncertainty
Model Uncertainty can be reduced with an appropriate DoE
Stage 3
An example: Vaccine compounding
Estimated concentrations
Titration
Decision:Proportion of
DS1 / DS2/ DS … / Buffer(% / % / … / %)
Drug Substance 1Buffer
Drug Product
Titration
Estimated concentration
Release Discard
Drug Substance 2
At mix After filtration At release At shelf-life
Overall view of the dilution problem
• Optimize the assay format, and the concentration of DS such that it will result in a drug product looking like…
- Each black line is the predicted behavior of one individual realization of DP
99% guaranteemeeting LSL atshelf-life…
LSL
Control strategy
• Control strategy is defined based on the (simulated) outcome of the process profile at strategic intermediate testing (red)
• The prediction interval (b-expectation tolerance interval) can be used as control limits.
Control strategy
• Raise appropriate out-of-control, alert, and reject at release
-30
-20
-10
0
10
20
30
Rela
tiv
e E
rror
100 (pg/m
l)
-30
-20
-10
0
10
20
30
Rela
tiv
e E
rror
1000 (pg/m
l)
-30
-20
-10
0
10
20
30
Rela
tiv
e E
rror
500 (pg/m
l)
Concentratio
n
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
series
It allows to control the risks and keep the quality constant over time.You maintain your initial claim and monitor it with appropriate levels of risk.
Release Routine
LSL
Conclusion
• Bayesian statistics provide a natural answer to all Stages of process or method development
• Bayesian statistics provide predictive distribution to permit prediction-based decision
• Prediction are key to Design Space
• Prediction are key to PPQ
• Bayesian statistics make multivariate modeling easy and allows to compute joint probability of success
• Bayesian statistics are easy to compute today with languages such as SAS, BUGS, JAGS, or STAN.
Thank you!