Efficient Risk Based Regulatory Scrutiny for Assuring Pharmaceutical Quality in the 21st Century

Download Report

Transcript Efficient Risk Based Regulatory Scrutiny for Assuring Pharmaceutical Quality in the 21st Century

Efficient Risk Based
Regulatory Scrutiny for
Assuring Pharmaceutical
Quality in the 21st Century
Ajaz Hussain, Ph.D. and David Horowitz, Esq.
CDER, FDA
Arden House 2004, London
Outline
 Challenging
opportunities for improving
efficiency and effectiveness

Science, Risk, Quality Systems, Harmonization
 Risk

based CMC & Biopharm Review
What type of “knowledge” may be most
useful
 Risk
based GMP Inspections
Part I: What type of “knowledge”
may be most useful for risk
based CMC Review
Evolution of Product Development
Data base -to- Knowledge base
Empirical models
Mechanistic models
Regulatory utility
Efficient and Effective Design

Product quality and performance achieved and
assured by design

Dosage form

Manufacturing process, ...

Specifications based on mechanistic
understanding of how formulation and process
factors impact product performance

Continuous "real time" assurance of quality

Highest (pragmatic) level of process understanding
Mitigate Risk
Risk of incorrect identity
Poor product & process
Changes in clinical trial product
Inadequate Design Specifications
Critical to quality and performance?
Risk of unqualified impurities
Risk of poor bioavailability
Risk of incorrect expiry date
Risk of inadequate controls
Risk of SUPAC,..
Risk of unrepresentative test
samples] Risk of Inadequate Facility
and QS
Tests & Controls -Risk Mitigation
Design for Intended Use: Focus on Reliability
Intended Use
Route of administration
Patient population
…..
Product Design
Design Specifications
(Customer requirements)
Regulatory
Specs.
Manufacturing Process
and its Control
Efficient Design: Different
Strategies
 Design
strategies are generally company
dependent
 Regulatory process should only focus on
the outcome of these efforts
 Certain information and knowledge utilized
during design justification and
development can be useful for science
based regulatory decisions
Regulatory Utility of
Development Knowledge
 Structured
knowledge that provides a
means for evaluating and predicting
product performance

Can be for certain aspects or be broad to
cover all aspects
 Fragmented/incomplete
not be useful

bits of data may
Need a general understanding and criteria for
the type of knowledge useful for regulatory
utility
Elements of Decision Making in PD
Long-term memory




Data - Information (communication)Knowledge -relationship in
Product Development
Significant reliance on
personal knowledge as
information source in PD
Generation of memory and
experience is achieved over
a long period of time
Court A W. Int. J. Info.
Management.. 17: 123, 1997
Short-term memory
Pharmaceutical Product
Development




Multi-factorial
complex problem
Significant reliance on
personal knowledge
Historical data likely
to have been
generated by a “Trialn-Error” approach
Many choices for
achieving target
specifications



Without up-to-date
information, high
potential for:
 misjudgments
 “reinventing the
wheel”
Mobile “institutional
memory”
Approved product
needs frequent
changes ..
Evolution of Product
Development Process
“Art”
Science & Engineering based
 Trial-and-Error
DOE
CAD
 Few creative options tested
Many creative options tested
 Dosage forms
Drug Delivery
Intelligent Drug Delivery
 CMC Review : Increased Reliance on
Pharmaceutics to optimize regulatory tests &
filing req.

Product Development
Knowledge
Level of Sophistication
HIGH
Details Resolved
Rules
HIGH
MECHANISTIC
MODELS
MEDIUM
LOW
EMPIRICAL
MODELS
HEURISTIC RULES
“Rules of Thumb”
HISTORICAL DATA DERIVED FROM
TRIAL-N-ERROR EXPERIMENTATION
MEDIUM
LOW
Artificial Intelligence (AI) and
Information Technology (IT) can
Improve the Utility of Historical Data
Expert Systems (AI)
Rules
MECHANISTIC
MODELS
Fuzzy ES
EMPIRICAL
MODELS
HEURISTIC RULES
“Rules of Thumb”
HISTORICAL DATA DERIVED FROM
TRIAL-N-ERROR EXPERIMENTATION
ANN
(IT)
Hypothesis (Hussain,1989): “ANN Based
Computer Aided Formulation Design”

Many Challenges




Very complex systems
Subjective descriptions
of excipient
“functionality”
Historic process
controls that do not
necessarily control
critical attributes of inprocess materials
Subjective equipment
similarity descriptions

Initial focus on



Comparison to RSM
Tool for improving
technical and marketing
support functions of an
excipient supplier:
Klucel Net  (Aqualon)
Formulation and
marketing tool for a
propriety formulation
technology: TIMERx
ANN Structure and Function
Data/Information
Ability to predict outcome
(Knowledge)
OUTPUT
INPUT
Simulation tool; “What-If” analysis
Pattern visualization tool, Optimization, etc.
Feasibility of Predicting Drug
Dissolution From Multisource
Tablet Formulations
V.K. Tammara,1 A.S. Carlin,2 M.U. Mehta1 and
A.S. Hussain 2
1
Division of Pharmaceutical Evaluation I, Office of Clinical
Pharmacology and Biopharmaceutics and 2 Division of Product
Quality Research, Office of Testing and Research, CDER, FDA
AAPS Annual Meeting 1999
Proposed SUPAC-IR Network
Phase III
Phase I
Phase II
In Vitro Dissolution
Drug attributes
and dissolution test conditions
Composition
Equipment (D/OP)
and Process
Methods
• Similarity between training (FDA/UMAB) and test (ANDA)
formulations
FDA/UMAB 1
2
3
4
5
6
7
8
9
Binder
  X
     X
Diluent A
X X
Disint.
        
Diluent B
        X
Lubricant
        
  X
   
Granulation         X
method
Test
Formulation
9
ANDA 1
Prediction
Error %:
Q(10)
-29
Prediction
Error %:
Q(30)
-15
ANDA 2
-2
-2
ANDA 3
13
13
ANDA 4
- 6
- 4
ANDA 5
25
4
ANDA 6
7
2
ANDA 7
14
-5
ANDA 8
- 4
4
ANDA 9
- 14
7
Innovator
6
-7
Results and Discussion
Component
Diluent A (IN)
Diluent A (OUT)
Diluent B
Disintegrant (IN)
Disintegrant
(Out)
Binder
Lubricant
SUPAC
Reco.
Limit
(Level 2)
10%
10%
10%
2%
2%
Max change in
component having no
impact on dissolution
(15% or less difference
in dissolution)
15%
30%
70%
7%
3%
1%
0.5%
5%
2%
Specifications

Product specifications based on mechanistic
understanding of how formulation and process
factors impact product performance

Continuous "real time" assurance of quality
Empirical Models: Experimental
Designs (e.g., Mixture Designs)
 Set
of combinations of the proportions (or
blends) used to collect observed response
values


range of interest of the formulator (entire
simplex surface or partial surface)
Properties of the polynomials used to
estimate the response function depend on the
experimental design
Factor Space
x1 =1
(1,0,0,0)
x1 =1
x1=x2=0.5
x2 =1
x1 =1
(1,0,0)
x4 =1
(0,0,0,1)
x3 =1
(0,0,1,0)
x2 =1
(0,1,0,0)
x2 =1
(0,1,0)
x3 =1
(0,0,1)
Response Surface
 Choice
of model to approximate the
surface over the “region of interest”
 Testing for adequacy of the model
 Suitable experimental design
 Assumption - existence of a continuous
functional relationship
   ( x1 , x2 ,....., x q )
Sathyan, G., Ritschel, W. A. and Hussain, A. S.: Transdermal delivery of tacrine, I:
Identification of a suitable delivery vehicle. Int. J. Pharm. 114: 75-83 (1995).
In Vitro Flux (Human Skin)
Permeability (Human Skin)
0.0 1.0
0.0 1.0
P
0.5
0.5
E
0.5
E
P
36
42
48
30
0.5
8
24
5
18
4
12
0.0
1.0
1.0
0.0
0.0 1.0
16
19
22
0.5
E
25
0.5
28
31
1.0
0.0
0.5
W
3
0.5
W
Lag-Time (Human Skin)
P
0.5
W
2
6
1.0
0.0
7
6
0.0
1.0
0.0
1.0
Drug-Polymer and Polymer #1Polymer #2 Ratios
 Effect
on drug release mechanism form a
hydrophilic matrix capsule containing

HEC + Na.CMC
 Drug
+ HEC + Na.CMC =1
 Variables:
M
 Response: n, T50% M  k  t
 Design: Factorial
Drug
HEC  Na.CMC
HEC
Na.CMC
t
n

Hussain, A. S., Johnson, R. D., Shivanand, P. and Zoglio, M. A.: Effects of blending a
nonionic and an anionic cellulose ether polymer on drug release from hydrophilic matrix
capsules. Drug. Dev. Ind. Pharm. 20: 2645-2657 (1994).
Optimal Formulations
Composition
Formulation 1
Formulation 2
HEC
0.35
0.43
Na. CMC
0.46
0.44
Drug
0.19
0.13
Response
Calc.
Obs.
Calc.
Obs.
n
1
1 (0.13)
1
1(0.08)
T50%
7.5
7.8(0.73)
8
8.1(0.8)
Region of Interest (Mixture+Process)
X1=1
Z=+1
w2
w1
X2=1
X3=1
Z=-1
w2
w1
X2=1
X3=1
A move towards “Mechanism”
(macro-scale)
 The
following slides provide a brief
summary of current research on topical
microbicide vaginal products

is an example of the “mechanistic” approach
• linking physics with physiology to identify critical
product attributes and explain how these attributes
effect product performance
Desired Distribution Profile of
Certain Vaginal Formulations
VAGINA
CERVIX
MUCUS
INTROITUS
PROPHYLACTIC COATING
CONTRACEPTIVE COATING
Interactions in the Vagina
Fluid
Contents
Anatomy,
Geometry
Microbicide
Formulation
Surface
Properties
Mechanical
Properties
Pre-Coital Forces Acting on a
Bolus of Gel in Vagina
“SQUEEZING”
visceral contractions
pressure
tissue elasticity
rugae
“SEEPING”
surface energies
interfacial tensions
“SLIDING”
gravity
David Katz. Duke University
Mechanistic Analysis of Subprocesses (Squeezing)
• THEORY
• SIMULATION
vehicle
epithelial surfaces
F
R
FORMULATION
2h
Area(t)=
David Katz. Duke University
Solution for elastic surfaces (E, n); lubrication
approximation; power law fluid (n, m); conserved bolus
volume V = 2hoπ R2
SLIDING THEORY
Vavg
velocity
profile
velocity depends upon…
vagina properties
tilt angle a
tissue separation Htot
gel properties
density
rheology
Htot
a
David Katz. Duke University
David Katz. Duke University
Axial and Angular Dependence
of Coating Thickness Distribution
Conceptrol
Advantage
Axial Dependence
Thickness (m m)
Thickness (m m)
Axial Dependence
1000
Scaled to length
of 71.11 mm
500
0
0
0.5
1000
500
315
270
500
0
225
135
180
% Volume Distal
to Fornix
1
Angular Dependence
Thickness in microns
85%
45
90
0.5
Axial Length
39%
0
1000
0
0
1
Axial Length
Angular Dependence
Thickness in microns
Scaled to length
of 51.67 mm
0
1000
315
% of Area Coated
36%
100%
270
500
0
225
45
90
135
180
http://www.fda.gov/cder/gmp/21stcenturysummary.htm
Desired State

Regulatory policies tailored to recognize the
level of scientific knowledge supporting product
applications, process validation, and process
capability

Risk based regulatory scrutiny relate to the:

level of scientific understanding of how formulation
and manufacturing process factors affect product
quality and performance, and

the capability of process control strategies to prevent
or mitigate risk of producing a poor quality product
Quality Risk Scenarios
 Risk

of unacceptable quality (examples)
Releasing a unacceptable quality product
• Inadequate controls/specifications


“New” impurities or degradation products
Bio-in-equivalence
• Inadequate process validation





Process not understood, sampling not “representative”
Stability failure
Unacceptable S&E profile, Bio-in-equivalence
Poor Process quality
Others
AAPS Poster, November 2001
STABILITY PROFILES OF DRUG PRODUCTS
EXTENDED BEYOND LABELED EXPIRATION
DATES
Jeb S. Taylor1, Robbe C. Lyon1, Hullahalli R. Prasanna1,
Ajaz S. Hussain2
Center for Drug Evaluation and Research, FDA, 1Division of Product Quality
Research,
Nicholson Research Center, Kensington, MD; 2Office of Pharmaceutical
Sciences, Rockville, MD
INTRODUCTION

The FDA administers the Shelf Life Extension Program
for the U.S. Military.

Selected drug products are tested to determine if their
shelf life can be extended past labeled expiration date.

Program probably contains the most extensive source of
stability data available.

The test attributes, methods of evaluation and analysis
are described.

This report summarizes extended stability profiles for
96 different drug products.
SUMMARY

Results from 1122 lots (96 drug products) were evaluated.

84% of the lots were extended for an average of 57 months past
the original expiration date.

Of the 946 lots extended, 14% were eventually terminated due to
failure. The rest are still active or discontinued by the military.

22 Drug Products showed no signs of stability failure (at least 5
lots of each tested).

10 Drug Products were unstable with most lots failing initial
extension.
Group 6: Unstable Products:
Most* Lots Fail Initial Extension
Drug Product
Albuterol
Brompheniramine Maleate Phenylpropanolamine HCl
Diphenhydramine HCl
Isoproterenol HCl
Levarterenol Bitartrate
Lidocaine HCl & Epinephrine
Mefloquine HCl
Penicillin G Procaine
Phenobarbital Sodium Inj
Physostigmine Salicylate
*  50% Occurrence
Dosage Form
Aerosol
SR Tablets
# of Lots # of Lots
Initial Extension
Tested Extended
Failures
2
0
Assay: Stress (2)
4
0
Appearance: Stress (4)
Spray
Injection
2
8
0
2
Injection
Injection
9
7
2
0
Tablets
Powder
Cartridge-Needle
Injection
19
5
3
14
4
1
1
4
Assay (2)
Assay (4)
Assay: Stress (2)
Assay: Stress (7)
Assay (2)
Assay : Stress (5)
Dissolution (15)
Assay (4)
Assay (2)
Low pH (10)
CONCLUSIONS

Actual Shelf Life may be much longer than indicated by
Expiration Date on the label.

Stability Period is highly variable from lot to lot.

Continued testing and systematic evaluation is required to
provide assurance that stability and product quality is
maintained.
YES
Dissolution
generally
“overdiscriminating”
Why?
Dissolution fails
to signal
bio-in-equi
~ 30% (?)
NO
Bioequivalent
Dissolution Test &
Bioequivalence: Risk
Assessment
NO
YES
Dissolution Specification
Appropriate Specification or “Overdiscrimination”
All Bioequivalent to RLD
DISSOLUTION OF GENERIC & RESEARCH TABLETS
110
100
90
ANDA1
ANDA2
80
ANDA3
ANDA4
70
ANDA5
60
ANDA6
ANDA7
50
ANDA8
40
ANDA8
74-217
30
UMAB-SLOW
UMAB-MEDIUM
20
UMAB-FAST
10
0
0
5
10
15
20
TIME IN MINUTES
25
30
35
% Drug Dissolved
Failure to Discriminate Between Bio-inequivalent Products: Inappropriate
Acceptance Criteria
Product B
110
100
90
80
70
60
50
40
30
20
10
0
Product B was not
bioequivalent to
Product A
Product A
Log(AUCinf): CI 94.6 - 123.6
USP Specification
0
10
20
30
Time in Minutes
Log(AUC): CI 89.1 - 130.0
40
50
Cmax: CI 105.3 - 164.2
Failure to Discriminate Between Bio-inequivalent Products: Inappropriate Test
Method?
(weak acid, rapid dissolution in SIF)
Capsule (Ref.)
1800
1600
Tablet 1
(wet-granulation - starch)
Drug Concentration in Plasma (ng/ml)
1400
1200
Tablet 2
(direct compression calcium phosphate)
1000
800
600
400
200
USP Paddle 50rpm, Q 70% in 30 min
0
0
1
2
3
Time in Hours
4
5
6
False Positives and False
Negatives!!!
Test/Ref. Mean
15 min 30 min 45 min AUC Cmax
Ref
95
96
98
100
100
B
96
97
97
104
95
C
62
84
92
84
55
D
82
94
95
88
87
E
103
103
103
112
120
F
13
35
53
100
102
I. J. MacGilvery. Bioequivalence: A Canadian Regulatory
Perspective. In, Pharmaceutical Bioequivalence
. Eds. Welling, Tse, and Dighe. Marcel Dekker, Inc., New York, (1992)).
Failure of Dissolution Tests to
Signal Bio-in-equivalence

Inappropriate “acceptance criteria”



Inappropriate test method




One point specification
Set “too late”
media composition (pH,..)
media volume
hydrodynamics
Excipients affect drug absorption
 Dissolution is NOT “RATE LIMITING”
 Other reasons (many critical test variables)
Is Dissolution Rate Limiting?
Concentration
3.5
Capsule
Solution
3.0
2.5
2.0
1.5
1.0
0.5
0.0
0
4
8
12
Time
16
20
24
ICH Q6A DECISION TREES #7: SETTING ACCEPTANCE CRITERIA
FOR DRUG PRODUCT DISSOLUTION
What specific test conditions and acceptance criteria are appropriate? [IR]
What?
YES
dissolution significantly
affect BA?
Develop test conditions and acceptance
distinguish batches with unacceptable BA
NO
Do changes in
formulation or
manufacturing variables
affect dissolution?
Why?
Are these changes controlled
by another procedure
and acceptance
criterion?
YES
YES
NO
NO
Adopt appropriate test conditions
and acceptance criteria without
regard to discriminating power, to
pass clinically acceptable batches.
Adopt test conditions and acceptance
criteria which can distinguish
these changes. Generally, single point
acceptance criteria are acceptable.
aaps Annual Meeting
How?
49
ICH Q6A: Decision Tree #7 (1)
Modified release?
Yes
No
Establish drug release
acceptance criteria:
ER: Multiple time point
MR: Two stage, parallel
or sequential
High solubility?
Yes
Rapid dissolution?
?
Yes
Relationship between
Disintegration - Dissolution?
No
No
Generally single-point
dissolution acceptance criteria
with a lower limit
No
Yes
Generally disintegration
acceptance criteria with
an upper limit
Disintegration - Dissolution Relationship
Disintegration
Time (DT)
Disintegration
Fraction
dissolved
Ft
Fl
Tablet
Surface (t)
10# screen
Fs
Large
Fragments (l)
Fs*
Small
Fragments (s)
Total Dissolution = Ft + Fl + Fs + Fs*
Prior to DT
After DT
Note: Disintegration and dissolution process in a dissolution apparatus may differ
from that in a disintegration apparatus (different hydrodynamics and other conditions)
Data
 Dissolution and disintegration time data and the impact of
formulation variables on dissolution at various time points (Figure 1)
C u m u la tiv e D is s o lu tio n a n d D is in te g r a tio n D a ta :
C r itic a l F o r m u la tio n V a r ia b le s
120
M C C (-)
S S G (+ )
[M g S (-)]
D r u g D is s o lv e d
100
M C C (-)
S S G (+ )
M g S (-)
80
60
%
40
20
0
C o r r e s p o n d in g
D is in te g r a tio n T im e
D a ta
0
5
10
15
20
T im e in M in u te s
Li et al, Pharm. Develop. And Tech. 1: 343-355, 1996 and
Rekhi et al., Pharm. Develop. and Tech. 2: 11-24, 1997
25
30
35
Predicting Dissolution of Experimental Furosemide Tablets
Using Near Infrared Spectroscopy
Everett H. Jefferson1, Alan S. Carlin1, Robbe C. Lyon1, Lawrence X. Yu2,
Robert L. Hunt1, Jin T. Wang1, William N. Worsley1, Ajaz S. Hussain3
1Division
of Product Quality Research, FDA, Kensington, MD
2Office
3Office
of Generic Drugs, FDA, Rockville, MD
of Pharmaceutical Science, FDA, Rockville, MD
Predicting Product Dissolution
Develop Casual Links based on Critical Attributes
Formulation-Processing
Casual
Link
NIR Spectra
Non-destructive Test
Casual
Link
Correlation
Dissolution
(opportunity to consider
direct link to PK/PD)
Destructive Test
Dissolution - Formulation: Casual Link
Dissolution is a function of processing variables:
Dissolution = f (Ex1, Ex2, P1, P2, PS…)
Ex1, Ex2 = Excipients (USP/NF)
P1, P2 = Process parameters (time, hardness ...)
PS = Drug particle size (specification)
Use Multiple Linear Regression (MLR) to determine the
variables critical to dissolution:
y = 0 + 1 x1 + 2 x2 + 3 x3 +
12 x1 x2 + 13 x1 x3 + 23 x2 x3 + ...
Dissolution-Formulation Correlation:
Direct Compression (%Diss at 15 min)
Regression Model: %Diss_15 = 56.3 + 25.3 x2 - 15.1 x1 x2
Filler
AcDiSol
MgS
Hardness
Filler*AcDiSol
Filler*MgS
Filler* Hardness
AcDiSol*MgS
AcDiSol*Hardness
MgS*Hardness
x2
x1 x2
The critical formulation variables (p<0.05):
disintegrant level
disintegrant and filler interaction
Within the design space only - e.g.,other variables held constant
NIR - Dissolution Correlation:
Direct Compression (%Diss at 15 min)
Predicted % Dissolution
100
Direct Compression (Lots 121-132)
PLS1 Model
2nd Derivative Spectra
80
60
Training Set (n=72)
R2 = 0.9771
40
Test Set (n=72)
R2 = 0.9465
20
0
0
20
40
60
Measured % Dissolution
80
100
NIR - Formulation: Casual Link
NIR Spectra of the Raw Materials
2.0000
3.0000
Furosemide
Avicel
Magnesium Stearate
2.5000
2.0000
1.5000
Intensity
Intensity
1.5000
1.0000
0.5000
1.0000
0.5000
0.0000
0.0000
-0.5000
-0.5000
-1.0000
-1.0000
-1.5000
1100
1300
1500
1700
1900
Wavelength
2100
2300
2500
Lactose Monohydrate
Ac-Di-Sol
Starch
Povidone
-1.5000
1100
1300
1500
1700
1900
2100
2300
2500
Wavelength
NIR - Formulation Correlations
Lactose
250
200
Direct Com pression
PLS1 Model
2nd Derivative Spectra
200
Training Set (n=72)
2
R = 0.9790
150
Avicel
250
Direct Com pression
PLS1 Model
2nd Derivative Spectra
Predicted Avicel (mg)
2.5000
Predicted Lactose (mg)
3.0000
100
Test Set (n=72)
2
R = 0.9786
50
0
Training Set (n=72)
2
R = 0.9741
150
100
Test Set (n=72)
2
R = 0.9742
50
0
0
50
100
150
200
250
-50
0
50
100
150
-50
Measured Lactose (mg)
Manufacturing
Constituent
Process
Ac-Di-Sol Direct Compression
Mg Stearate Direct Compression
Starch
W et Granulation
Povidone
W et Granulation
Measured Avicel (mg)
Training Set
R2
n
72 0.9897
72 0.9780
48 0.9758
24 0.9911
Test
n
72
72
48
24
Set
R2
0.9860
0.9781
0.9539
0.9873
200
250
Process
Understanding
CMC
regulatory
oversight
cGMP
regulatory
oversight
Company’s
Quality system
Post
approval
change
Risk
(P/R)
Process
Understanding
CMC
regulatory
oversight
Process
Understanding
CMC
regulatory
oversight
cGMP
regulatory
oversight
cGMP
regulatory
oversight
Company’s
Quality system
Company’s
Quality system
Post
approval
change
Risk
Post
approval
change
Risk
CMC-GMP Risk Model

Should allow us to recognize the level of process
understanding for a given product
 Provide a means to move towards a one time
CMC review process for well understood
products and processes (current system will
remain for others)



Exception - unless specifications need to change
Continuous improvement within a companies quality
system
Share and have available knowledge throughout the
organization
Risk Mitigation Opportunities
•
Advances in application of risk analysis and
risk management (including IT revolution)
• More systematically applied outside
pharmaceutical sector

Including data generated by advanced manufacturing
technologies
• Risk management approaches to regulatory
compliance gaining wider acceptance among
regulators
Risk to Quality
 “Risk”
to pharmaceutical quality is
associated with factors that relate to the
probability or severity of adverse effects
on these attributes

Factors associated with compromising :
• e.g., identity, strength/potency, bioavailability,
purity, or clear/accurate labeling
Applying Risk Management to Drug
Quality Regulation (cont’d)
 Tools
and techniques for identifying and
focusing on critical processes and
parameters potentially include:




Failure Modes and Effects Analysis (FMEA)
Hazard Analysis and Critical Control Points
(HACCP)
Fault Tree Analysis (FTA)
Hazard and Operability Study (HAZOP)
Applying Risk Management to
Drug Quality Regulation (cont’d)
 To
support more focused regulatory
scrutiny we need:

Greater understanding of the sources of risk
to product quality
• Including factors predictive of that risk, as well as
those that are predictive of the successful
mitigation of the risk; and

Enhanced hazard identification and risk
assessment capacity
Applying Risk Management to Drug
Quality Regulation (cont’d)
 Greater
uncertainly requires greater
regulatory scrutiny


Burden must rest largely with regulated
industry to demonstrate to regulators that
reduced regulatory scrutiny is justified by the
science/data
Regulators need to support industry efforts to
develop and provide the needed science/data
Applying Risk Management to Drug
Quality Regulation (cont’d)
 Risks
to pharmaceutical quality can be
ranked by identifying and weighting factors
associated with the probability or
severity of adverse effects on quality
attributes and surrogates


What processes and parameters are critical for
identity, strength/potency, bioavailability, purity, or
clear/accurate labeling?
What factors will adversely (or positively) affect critical
parameters/processes, increasing probability or
severity or impact?
Risk Management and
Resource Allocation

FDA is developing a model using a Risk Ranking
and Filtering technique to better focus
inspectional resources based upon the risk of
manufacturing deficiencies that would reduce
drug quality


systematically incorporate our current knowledge
about drug quality risks and manufacturing sites
Prioritize sites for periodic systems-based inspections
Risk Management and
Resource Allocation (cont’d)
 Starting
in FY ‘03, shifted emphasis to
facilities making drugs perceived to be of
higher risk if subject to manufacturing
deficiencies
• Sterile drugs
• Rx drugs (non-medical gas)
• New registrants
 New
model under development for pilot
implementation later this year
A RISK-BASED FRAMEWORK FOR PRIORITIZING SITES FOR cGMP
INSPECTION
SITE RISK POTENTIAL
PRODUCT
PROCESS
Site potential risk score:
[wp Product weight] x [wPr Process weight] x [wfx Facility weight];
“w” is the relative weight assign to the individual module
Semi-quantitative weights
high, medium, low; yes/no; ordinal scales 1- 5
FACILITY
Risk Ranking Model:
Product Factors

What are the intrinsic properties of drug products
such that deficiencies in quality, if any, would
have more adverse public health impact than
others?




sterile
Rx vs. OTC
Route of administration
Mining recall data can help weight product
factors (e.g., product or dosage form associated
with prevalence of serious recalls?)
Risk Ranking Model:
Facility Factors

Are some manufacturing facilities (or
manufacturers) more likely to produce a product
with quality problems?


Effectiveness of quality systems and process
capability
Inspectional record and compliance history
• Exposure: volume produced at facility


Product sales volume
Special/sensitive populations
• Other characteristics?


New Registrants?
Macher and Nickerson study will help identify
Risk Ranking Model:
Process Factors

Are some manufacturing processes, for
particular product classes, more likely to go
wrong than others?

Are some process problems of greater public health
significance? What are the consequences of process
problems?
• Use expert elicitation to identify risk factors and weightings


At the unit operation level
By product classes
• Risk of contamination or mix-ups
• Maintaining state of control of the process
Risk Management and
Resource Allocation (cont’d)
 Risk
ranking and filtering technique may
be a useful tool for other aspects of drug
quality regulation




Focus compliance programs?
Guidance development?
Regulatory action decisions?
Industry internal audits?
Risk Management and
Resource Allocation (cont’d)
 The
product can only be as good as the
scientific/technical input/assumptions that
are used to develop the risk scores
 Multiple iterations and successive
revisions will reflect growing knowledge
base and extensive input from internal and
external experts