Analytical methods Quality by Design (AM-QbD) Risk Assessments undertaken within

GlaxoSmithKline (World Wide Inhaled Product Development) for Orally Inhaled and Nasal
Products
Andrew J Crumpton, Geoffrey E Daniels, Karl M Ennis and Nick C Turner. With Acknowledgements to Andrew J Rice.
GlaxoSmithKline R&D, Ware, Hertfordshire, UK

Introduction Process Map linked to DFM

Fundamental to any method development is being clear on the design intent of the method. Method Performance Criteria Inputs
Process Map
and Method Operational Intent are two important aspects of this design intent. Method targets for selecti vity, s ensiti vity,
precision, etc
Design
Intent

Project Background
Voice of the Customer
Method development s trategies
1 Method
MethodDDevelopment
evelopment Design
Outputs Selection
A fit for purpose method
Throughout the lifecycle of a given Orally Inhaled and Nasal Drug Product (OINDP) method it is inevitable that there will be 2 Risk
RiskAssess
Assessment
ment
Identify the
design space

Param eters
changes in the ‘method environment’ that can impact its operation. Changes/improvements to a method should be made with

which can
Inputs

be fixed
Inputs
Parameters
Noise Factors selected
selected for study
for study
reference to a method knowledge repository. GSK’s repository is entitled EMPACT (Experimental Method Parameter 3 Method
MethodRobustness
Robustness 4 Method
MethodRuggedness
Ruggedness
Control
Definition
Assessment Control Tool).
Aims
5 Control
ControlDefinition
Definition

Method Method
6Validation
Validationand
andus
usee

Ø Describe the repository and provide an overview of the analytical method risk assessment process used at GSK for
Control
Continuous
ContinuousImprovement
7 and
andCChange
Improvement
hangeControl
Control
Verification
Outputs
OINDPs with some examples for MDI and DPI Products. Learnings
FME A R is k
V a ria ble Pr oc es s PM Im por ta nc e PM R isk Sc or e FM EA R is k
V ar ia ble

C a rto n in g to p ro tec t o ve rw r ap - o ve rw ra p pe d pa ck s
C la s s S te p Sc ore (IS ) ( H /L )
Pr iority N um be r
(R PN )
Sc ore (L/M /H )
Inclusion of
E2
E1 r ec e ive d in b ox es of 5 0 - ne e d to b e r an d om i se d p ri o r to
s tab i li ty .
C r ac ke d foi l
C

N
0

0
0

40 M
factors to be
A 12 0 0

INSTR UCTIONS &

HELP Pag e
A 10
A 14
B2
H o w qu i ck ly d oe s e q u il ib ra tio n ta k e p la ce w ith a g i ve n
X
0
0
8 H
0
0
0
Controlled
p a ck wi th a g iv en p ro be .

SETUP &
INFORM ATION Pag e
B4
D 20
C4 0 0
0
0
0
0
0
0
alongside
C6
C7
C8
0
0
0
0
0
0
‘’X’’ and ‘’N’’
C1 Q ua l ite ch w ill on l y me a su re h ol e s >25 m ic ro n 0 0

Patient Requirements VARIABLES SCORING & CONTROL STUDY SELECTION QbD TOOLS & REPORT TEMPLATES
C2
C3
F or p ro b e me a su re m en t w e n e ed to m i ni m is e l ab a ir
i n gr es s i nto ov e rw ra p
V is u al in sp e ctio n
C
C
0
0
0
0
factors for a
A METHOD
Method Performance F2
F3
B7
C o n si ste nc y o f te sti ng
P o si ti o n/s ize o f h o le fo r pr o be
C
C
0
0
0
0
0
0 DPI product
ROBUSTNESS
B
EQUIPME NT

ENVIRO NME NT
EXPERIMENTAL (X)
METHO D DEVELOP MENT/
PROCE SS SCOPI NG
STUDIES ( DoE 1)
DEVELOPMENT
REPORT REPORT
STUDIES Requirements B3
B6
F1 D a m a ge to p a ck du ri ng h a nd li n g C
0
0
0
0
0
0
VARIABLES

Process Design & C MEAS URE ME NT PRIORITISATION M ATRIX METHOD/PROCESS

REPORTING
RUGGEDNESS
REPORT
ROLLED
FUNCTIONSW
STUDIES WALK
ILL BE
THRU
OUT IN A LATER VERSION
Occurrence Risk Priori ty Ri sk Score
Include i n
D Variable Fai lure Mode (Category) Fai lure Effects Severity (S)
(O)
Detection (D)
Number (RPN) (L/M/H)
Ruggedness

Development METHO D
ROBUSTNESS
Study

E
MATE RIALS STUDIES (DoE 2) Method Development A2 Mouthpiece EQUIPMENT
poor fit resulting in leak/f low rate, angle
change when firing into s tack
4 1 2 8 L

F PEOPLE
NOISE (N)
VARIABLES
CONTROLLED (C)
EMPACT SUMMARY

VARIABLES REPORT
REPORT A3
On t he fully qualif ied Site 2 Automated Systems, 3 sets of 5 CIs each by 3
analysts (tot al 15) will be collec ted bot h manually and using t he Automat ed
Sys tem. The testing must be c omplet ed within 1 month for the sample
data sets to be equivale nt.
EQUIPMENT
dif ferent timings from diff erent units and
dif ferent configurations lead to failed s pec .
5 5 5 125 H YES Example of
FMEA

Risk Assessment & Process RUGGEDNESS FISHBONE MINDMAP VARI ABLE LIST A6
A4 s tack to sample alignment

t emp/humidity c ont rol

EQUIPMENT

EQUIPMENT
poor fit resulting in leak/f low rate, angle
change when firing into s tack
repeat t est
3

1
2

5
1

1
6

5
L

L
‘’N’’ factors
STUDIES (MSA) Risk Assessment & Analytical A7 s ample to f ire motor alignment EQUIPMENT poor repeatability, st ep change 5 5 1 25 M Possi bl e

Design Space Definition GENERI C MODE PARETO C HA RT PM CHARTS Method Design Space Definition
A8
A9
A10
A11
s ystem t imings (bet ween syst ems)
differences bewteen prime and collec tion (int er and int ra build)
s hak e to fire delay differences (different builds)
EQUIPMENT
EQUIPMENT
EQUIPMENT
EQUIPMENT
poor repeatability, st ep change
poor repeatability, st ep change
poor repeatability, st ep change
5
5
5
5
5
5
1
1
1
25
25
25
25
M
M
M
M
Possi bl e
Possi bl e
Possi bl e
Possi bl e
held in
differences in met hod settings (between sites) - release speed. poor repeatability, st ep change 5 5 1

SPECIFIC MO DE
CONTROLLED (C)
VARIABLES VARI ABLES ASSO CIATE D REPORTS PROCE SS MA PS
A12
A13
A14
differences in s ample holders between sites
differences in pumps (DC v s AC)
airflow at the was te position between unit s
EQUIPMENT
EQUIPMENT
EQUIPMENT
poor repeatability, st ep change
poor repeatability, st ep change
poor repeatability, st ep change
5
5
5
5
5
5
1
1
1
25
25
25
M
M
M
Possi bl e
Possi bl e
Possi bl e
EMPACT for
A16 Ram tips - s tandard mat erial EQUIPMENT 1 1 1 1 L
WITH METHO D/PRO CESS
View Scor es
STEPS
Analytical Method Control
B3

B6
Room being designed at Site 3 wit h control on Temp +/- 3C but no humidity
c ont rol
Pump warm up time
ENVIRONMENT

ENVIRONMENT
delay in priming and collec tion, going out side
the method range during collection
unstable f low rate resulting in change in CI
5

5
4

1
1

5
20

25
M

M
Possi bl e

Possi bl e
an MDI
Control Strategy Hide Sco re s

Strategy B7 posit io n of temp/humidity probe ENVIRONMENT

prof ile

outside of method parameters (may result in

testing when outside parameters or s topping 5 4 1 20 M Possi bl e product
test thinking outside the parameters)
B8 acet one rinse impact ENVIRONMENT low res ults on 6,7F (Product A) 1 1 1 1 L
B9 s tack equilibration in the Automated System ENVIRONMENT dif ferent particle size distribution 5 3 1 15 M Possi bl e
B10 locat io n of the Automat ed Sys tem ENVIRONMENT 1 1 1 1 L
D9 I nnova for fire down METHOD dif ferent BOU t o EOU rise 3 3 1 9 L
F2 Analyst test ing PEOPLE not applicable to Automat ed Sys tem 0 0 0 0
F3 Analyst test ing (seating shot before putting on Automated System) PEOPLE af fect force to actuate during learning s tep 5 2 1 10 L

Im port ance Risk Score

Inclu de in Method
Include in
Example of
Variable Accuracy Precisio n Repeatability Reproducibil ity Selectivit y Lin earit y Development
Score (IS) (H/L)
(YES)
Robustn ess Stu dy ‘’X’’ factors
Method Performance Criteria held in
QbD for Inhaled Product Automation A5
A15
instr ument vibration 5 5 5 5 1 1 22
12
H
H
YES
YES EMPACT for
Damping system dif ferent from site 1to site2 1 3 3 3 1 1
Method Performance Criteria are driven by an B1 look at impact on time betweenprimingandcollection 5 5 5 5 1 1 22 H YES
• Cross inhaled Project D1 Shake duration 5 5 5 5 1 1 22 H YES YES
an MDI
Equipment Implementation – 1st Machines understanding of the process monitoring and control D2 Shake Speed 5 5 5 5 1 1 22 H YES product
application D3 Hold time 5 5 5 5 1 1 22 H YES YES
Manual Automated requirements and the ability to maintain the Product Design D4 Speedto actuate 5 5 5 5 1 1 22 H YES
(automation MDI and D8 Releasespeed 7 5 5 5 1 1 24 H YES
Space, i.e. the ability to meet the required critical quality
In d iv id u a l M a c h in e

URS URS URS Define the requirements

DPI)
Outputs from Prioritisation matrices and FMEA
Q u a lif ic a tio n

NGI NSRS NDSI

attributes (CQAs). CQAs are identified, through a thorough
understanding of those attributes of a drug substance or a exercises that come out of AM-QbD Risk
• QbD example for IQ IQ IQ Correctly Installed
drug product which may need to be controlled in order to Assessments held in EMPACT templates.
introduction of inhaled OQ OQ OQ Controls work correctly
assure the safety or efficacy of a product. This control can Any proposed changes that
product automation Performs the functions
30
5
PQ PQ PQ be achieved via specification limits, In Process Controls, in take the method outside its

Risk P riority Num ber (RP N)

25
has been presented to as required
4
Method Guidelines
Defines the set-up and use line monitoring and Process Analytical Technology etc. 20
proven design space are risk
I n d iv id u a l

FDA
D e ve lo p

Occurrence
P ro d u c t

M e th od

a) NGI User Guide 3

b) NSRS Method Guide Method Performance Criteria are defined by using a 15
assessed and for high risk
Method Validation Validation of the method rigorous approach for identifying all the potential method 10
2
changes an evaluation or
• Reduction of 5 1
equivalence exercise is
factors that need to be controlled to assure method
T est U se

Assesses the variability

S y s te m

Measurement System Analysis

potential Analyst error of the test systems 0 0 performed to assure method
performance and through the use of risk assessment tools

F3

F2
B6

B7

B3

A2

B8
using Automation Implement Globally System is valid for use
Variables
performance criteria are still
and prioritised experimentation that eliminate areas of risk. met. Method specific learnings
Semi-Automated NGI DPI Sample Solution Preparation
12

Blister Strip Apparatus (BSA) 10 are used to update the method

• Manual manipulation of the strip 8 knowledge repository.

FMEA Score
• Controlled alignment of piercing 6 Technique learnings are used
• Standardised process 4
to enhance the risk
• Sample recovery by pumped solvent 2
assessment process for future
• Volumetric process 0
methods. This is particularly

F3

F2
B6

B7

B3

A2

B8
• Removal of wash bottles Var iables important for automated

Conclusion
Variable Failure Mode (Category) Fail ure Effects Severity (S)
Occurrence
(O)
Risk Priority Risk Score
Detection (D) Number (RPN) (L/M/H)
Include in
Ruggedness methods for OINDPs.
Study
A8 Air bubbles in line EQU IPMENT Reduced solvent possibly poor rec overy 1 2 4 8 L
A9 Piercing speed EQU IPMENT possibly blow blend out into volumetric 2 3 5 30 M Possible
A12
A16
A18
Position of s trip in groove
Blunt needle
Tubing
EQU IPMENT
EQU IPMENT
EQU IPMENT
potent ia lly low recovery
not pierce properly
Reduced solvent possibly poor rec overy
5
2
1
2
1
2
1
4
4
10
8
8
L
L
L
As demonstrated, the AM-QbD Risk Assessment process
B1 Temperature ENVIRONMENT Inaccurate solvent volumes 2 1 2 4 L
B2
B3
Position of equipment in lab
Vibration of adjacent equipment
ENVIRONMENT
ENVIRONMENT
no known eff ects
no known eff ects
0
0
for OINDPs has been fully considered to fully mitigate the
B4 RH ENVIRONMENT no known eff ects 0
D2
D3
E1
Drip / draining time
Blis ter marking (e. g. pencil, no mark in g, etc.)
Blis ter dimensions (e.g. depth, width)
METHOD
METHOD
MATERIALS
low recov ery
contamination and inc orrect id of blisters
not pierce properly
2
2
5
1
2
2
5
2
1
10
8
10
L
L
L
analytical risk associated with testing OINDPs in different
E2 Sealing of s trip MATERIALS
Was h the landfill area around t he blister
(within the clamp)
Shallow needle was h height, not washing
2 1 5 10 L
laboratories within GSK. Key to this process is the scoring
E3 Foil thickness MATERIALS 2 1 5 10 L
E5
F1
Solvent viscosity
Change Height of platform
MATERIALS
PEOPLE
properly, potential low recovery
no known eff ects
Potential los s of drug with splash back 3 2 4
0
24 M Possible
system (for prioritising potential method factors that need
F2 Speed of levers PEOPLE Same as A9 0
F3
F4
Non tightening of platform
Strip placement
PEOPLE
PEOPLE
Fla sk can fall and break
Same as A12
5 1 1 5
0
L
to be mitigated) that has been developed, piloted and
F5 Incorrect lev er proc ess PEOPLE No s olvent into flask, no recovery 5 2 1 10 L
F6
F7
not clamping red lever completely
Forgetting to pierce
PEOPLE
PEOPLE
Solvent leak , poor recovery
Solvent leak , no rec overy
Pierced by solvent force, potential low
5
5
2
2
1
1
10
10
L
L established within GSK R&D.
F8 Incorrect lev er proc ess PEOPLE 2 2 4 16 M Possible
recovery

Corresponding author: Andrew J Crumpton. World Wide Inhaled Product Development, GSK, R&D Created by Information
Design