1. Manufacturing, Measurements & Quality Systems: Seeking Order in Complexity
Richard L. Friedman
Director,
Division of Manufacturing & Product Quality
Center for Drug Evaluation & Research
SLIDE 29 – Revised to remove redundancy

2. ICH Q9 Risk Control Review Events Risk Review Initiate
Risk Assessment
Risk Evaluation
unacceptable
Risk Control
Risk Analysis
Risk Reduction
Risk Identification
Review Events
Risk Acceptance
Initiate
Quality Risk Management Process
Output / Result of the
ICH Q9 Risk Control
Risk Communication
Risk Management Tools

3. Pharmaceutical Manufacturing
“Conventional pharmaceutical manufacturing is generally accomplished using batch processing with laboratory testing conducted on collected samples to evaluate quality.”
“…significant opportunities exist for improving
development, manufacturing, and quality assurance
through innovation in product and process development, process analysis, and process control.”
Guidance for Industry: PAT – A Framework for Innovative
Pharamceutical Development, Manufacturing, and Quality Assurance (2004)

4. The Quality System: Foundation for Assuring an Ongoing State of Control
Includes:
Materials System
Equipment & Facilities
Production
Laboratory
Packaging & Labeling
Quality System

5. Adaptation: Process Control and Improvement
FEEDBACK SYSTEMS:
“Control procedures shall be established to monitor the output and to validate the performance of those manufacturing processes that may be responsible for causing variability in the characteristics of in-process material and the drug product” ( a)
IMPROVEMENT:
Firm must review, at least annually, the quality standards of each drug product to determine the need for changes in drug product specifications or manufacturing or control procedures. ( )

6. Goal of a Manufacturing Organization
Provide a consistent, defect-free product to the marketplace via consistent manufacturing operations
Assure safety & efficacy
…But not every company has established adequate quality practices, or achieved predictable output. Some companies, products, or processes make more mistakes /defective units than others. Why?

7. “Putting out fires is not improvement of the process”
- W. Edward Deming

8. Each production day, each dose, each patient
If you are operating at 3.8 Sigma, you are getting it right 99 percent of the time... It turns out that even a 1 percent error can add up to a lot of mistakes pretty fast. Getting it right 99 percent of the time is the equivalent of 20,000 lost articles of mail every hour. It’s 5,000 botched surgical procedures every week. It’s four accidents per day at major airports...
If you can answer when, where and how often the defects occur you have what you need... But don’t just focus on the symptoms of the problem. Find the root causes.
Chowdury, 2001

9. Current state of Pharmaceutical Manufacturing
Operations in Pharmaceuticals Compare Poorly to Other Industries
The pharmaceutical industry lags similar industries in key measures of operations performance…Many of the shortcomings reflect poor quality practices and represent cost savings opportunities... Estimates are from McKinsey Operations Practice.
Measure
Pharma
Automotive
Aerospace
Computer
Consumer Packaged Goods
Overall equipment effectiveness
10% to 60%
70% to 85%
50% to 70%
80% to 90%
70% to 90%
Annual productivity improvement
1% to 3%
5% to 15%
5% to 10%
First-pass yield - zero defects
60%
90% to 99%
Production lead times in days
120 to 180
1 to 7
7 to 120
5 to 10
3 to 7
Finished goods inventory in days
60 to 90
3 to 30
5 to 50
10 to 40
Labor value-add time
20%
60% to 70%
60% to 90%
Direct/indirect labor ratio
1:1
10:1
The Gold Sheet, January 2009

10. Why Measure?

11. “The basis of all scientific work is the conviction that the world is an ordered and comprehensible entity”
– Einstein 11

12. For a Measurement to provide meaningful evaluation of quality, it should:
Provide a direct measure of the critical in vitro attribute that is the surrogate for the in vivo clinical attribute (ideally with IVIVC)
Provide very high assurance that process reproducibly meets its limits & specifications for each batch
Have a purpose (why measure?)
to detect and prevent non-conforming in-process and finished product output – i.e., assure effective control of the process
…and thus greatly reduce risk of any defective product reaching the consumer
Be sufficient and timely (how, where, when & how often?)
Be reliable. Measurement capability (good analysis provides the truth; allows for correct decisions) 12

13. How, where, when and how often to measure?
Tests
Qualitative vs. quantitative
Instrument and operator interfaces
to process
to sample
Timeliness and frequency
Method of sampling
destructive or non-destructive
at-line (sample removed), offline (sample temporarily diverted off-stream) or in-line (on-stream from a distance)
invasion of process stream can affect measurement
representative of the batch or providing worst case assessment
periodic, continuous, stratified to risk 13

14. Sources of Variability
When developing or evaluating a process, essential questions include:
What variables in the process are influential, and require careful control and intensified monitoring?
Why and how can a particular variable affect final product quality?
How should cGMP procedures be designed to control that variable? Each significant variable can be measured in one or more ways. As variables are measured, a reliable picture of batch quality will emerge.

15. How Good Are Our Measurements?

16. Quality System Detection of Variation & Defects before Distribution
Test of a firm’s Quality System is if it will promptly catch a problem in a batch vs. discovering only after it is marketed.
Mistakes are, in many cases, not caught by the individual making the error, but instead through final inspection or QC test!
QC testing is of limited sample size intended to assess a chemical, microbiological, or physical attribute.
To avoid detecting mistakes or defects only after a drug product has been distributed:
Use Redundancy of Controls, or PAT

17. Product Quality: How good is batch-to-batch QC testing?
Specifications alone can not assure product quality.
Quality must be built in (21 CFR §§ , )
QC testing regimens may be found during lifecycle to be insufficient. Enhanced process control or new/modified QC tests might be needed.
Passing result for finished product potency, dissolution or content uniformity, on its own, does not provide high assurance that it will pass again.
Sample sizes are not always sufficient, and processes are not always stable.

18. Potential for Dissolution Variability
If a given lot just barely passes the USP or approved criteria once, what is the probability of passing 2 to 8 more times?
Depending on the overall inherent lot variability, it can range from 0.3% to 100%

19. Case Study #1 Dissolution Failure (multivariate cause?)
It was concluded that the cause of the dissolution failure was a combination of factors, including a formulation change, specifically a 1% increase in lubricant, a subtle change in the effective density of the tablets, the bulk density of microcrystalline cellulose and the process which changed mixing principle (v-blender vs. shaker mixer).
Dissolution testing on stability ultimately detected the problem.
3 batches failed (but only 1 batch/year typically placed on stability by firms)

20. Case Study #2 Subpotency (multivariate cause?)
Tablet product
Assay failure on Stability (9 months)
Firm’s investigation concludes that product stability needs to be improved by:
changing to a different API source
modification of formula
improved container/closure system

21. Microbiological Measurements: Complexity of Measuring Risk to the Patient
Type of organism
“Opportunistic not primary pathogens cause most infection”
Infective dose
Very elusive based on next factor…
Host resistance/susceptibility to infection
Route of Administration
“In general, the risk of infection will be much reduced for a drug given orally or applied to intact skin compared with a formulation used for treatment of abraded skin or mucous membrane, or damaged eye.”
[Bloomfield, “Microbial Contamination: Spoilage and Hazard.” Chapter 2 in Guide to Microbiological Control in Pharmaceuticals, 1990.]

22. Interpreting Atypical Measurements: OOS Guidance
3/31/2017

23. OOS Guidance
3/31/2017
FDA’s 2006 Guidance addresses:       Conducting/concluding the investigation 2.      Interpretation of results 3.      Handling of inconclusive results 4.      Retesting 5.      Appropriate use of averaging 6.      Appropriate use of outlier tests

24. 3/31/2017
OOS Guidance: General
Recommended procedures for OOS investigations are divided into two phases to reflect that the OOS result can be caused by either:
An aberration of the measurement process (i.e. laboratory error)
An aberration of the production process (i.e. the product is OOS)

25. OOS Guidance: General
3/31/2017
Too often root cause is unknown, or arbitrarily attributed to an analyst. Two possibilities:
Product related variability: Formulation components, manufacturing process, operator, etc.
Measurement system (method) related variability: Sampling bias, apparatus setup, analytical, operator, reference standard, etc.

26. Warning Letter – Data Integrity Can You Trust the Reported Measurements?
Failure to establish appropriate controls over computer or related systems to assure that changes in master production and control records or other records are instituted only by authorized personnel (21 CFR § (b)). The UV/Visible spectrophotometer data acquisition systems allow analysts to modify, overwrite, and delete original raw data files.
This equipment is used for dissolution testing of finished product, stability samples, and process and method validation studies.
All laboratory personnel were given roles as Managers, which allowed them to modify, delete, and overwrite results files.
This system also does not include an audit trail or any history of revisions that would record any modification or deletion of raw data or files.
Your laboratory computer system lacks necessary controls to ensure that data is protected from tampering, and it also lacks audit trail capabilities to detect data that could be potentially compromised.
[Paraphrased]

27. OOS Guidance - Investigations
OOS results may indicate a flaw in product or process design.  For example, a lack of robustness in product formulation, inadequate raw material characterization or control, substantial variation introduced by one or more unit operations of the manufacturing process, or a combination of these factors can be the cause of inconsistent product quality.  In such cases, it is essential that redesign of the product or process be undertaken to ensure reproducible product quality.

28. Building knowledge…Opportunities for Variability Reduction and Innovation
“Implement appropriate product quality improvements, process improvements, variability reduction, innovations and pharmaceutical quality system enhancements.
[ICH Q10 ]

29. NIR in PAT
Provides NIR spectral signature (reflectance from solid or transmission through liquid) with adequate signal/noise ratio
spectra contain information on chemical composition as well as physical attributes of particulates
Process Endpoints: information can be used to monitor progress of a chemical or physical change, and determine unit process endpoints
API synthesis, drier moisture endpoint, blending, solvent evaporation, online content uniformity and tablet weights, etc.
Increasingly used for real time release

30. Development Predictions30

31. Steps in Pharmaceutical Process Development & Scale-up
Validation of
Predictions
Predictions using models
Experimentation and Data Analysis
Learning

32. Building knowledge…Development
“Monitoring during scale-up activities
can provide a preliminary indication of process performance and the successful integration into manufacturing. Knowledge obtained during transfer and scale-up activities can be useful in further developing the control strategy.”
[ICH Q10 – Technology Transfer]
Monitoring during
Monitoring during
scale-up activities
scale-up activities
can provide a
can provide a
preliminary
preliminary
indication of process
indication of process
performance and the
performance and the
successful integration
successful integration
into manufacturing.
into manufacturing.
Knowledge obtained
Knowledge obtained
during transfer and
during transfer and
scale up activities
scale up activities
can be useful in
can be useful in
further developing
further developing
the control strategy
the control strategy

33. From Development to Commercial Scale
Process Development predictions can be affected by:
Lack of fundamental knowledge and understanding of scale dependent factors
Lack of correspondence between equipment used in lab and in factory
Lack of full understanding of critical attributes of raw materials, and how they affect physics and mechanics of processing
Finite time lines and pressures to market curtail some process development and scale-up activities

34. Predictions - DOE Weaknesses
Based on model
Predictions are extrapolations
inside as well as outside explored space
Experiments done at lab scale
Missed factors
Missed interactions at screening
Each factor alone has little impact
e.g., cycle # and regeneration buffer salt
Greater weakness for complex products and processes
Dr. S. Kozlowski, 2008 34

35. Pilot Scale and Lot Data
Statistical Process Analysis
Multivariate SPA
Gap between process experience and DOE
Adapted from T. Kourti 35

36. [Jean-Marie Geoffroy, May, 2007]36

37. Building Knowledge… Monitoring and Adaptation
Pre-approval
Process performance and product quality monitoring systems for each product should include:
Management review of process performance and product quality
Corrective action & preventive action (CAPA) and Change Management
Continuous Process Verification (CPV ) provides strong scientific basis for maintaining “state of control” throughout commercial lifecycle.
Post-approval
Product Lifecycle 37

38. “Constantly and forever improve the system of production and service”
- W. Edward Deming

39. CGMP: Every batch, Every day…
“We rely upon the manufacturing controls and standards to ensure that time and time again, lot after lot, year after year the same clinical profile will be delivered because the product will be the same in its quality… We have to think of the primary customers as people consuming that medicine and we have to think of the statute and what we are guaranteeing in there, that the drug will continue to be safe and effective and perform as described in the label.” - Janet Woodcock, M.D. 39

40. Summary: Paradigm Shift
Quality Should Be:
Proactive, rather than reactive
Built on knowledge and understanding of product, process and materials
Built in through design of facilities, equipment and processes (CGMP)
Based on use of state of the art tools of measurement science to control processes
Utility of Better Measurements in Quality Assurance
Improved process control
Continuous quality monitoring in real time
Automated, objective decision algorithms
Data can be analyzed retrospectively (performance monitoring) to trigger improvements in process capability

41. Acknowledgments: For More Information: Tara Gooen Vibhakar Shah
Lynn Torbeck
Steve Wolfgang
For More Information: