quality by design (qbd) solutions for analytical method ......8 qbd approach for method development...
TRANSCRIPT
Quality by Design (QbD) Solutions for Analytical Method Development
A Systematic Approach to
Reducing Variability
Content
• Introduction- Traditional Development and Transfer of Methods
- QbD Approach for Method Development
• Agilent Solutions for QbD Method Development-Agilent Method Development Systems
-Intelligent System Emulation Technology (ISET)
• QbD Method Development Workflow- Screening
- Optimization
- Robustness study, Design of Experiments
- Transfer & Verification
3
Key PointsQuality by Design (QbD) Solutions for Analytical Method Development
• Agilent 1290 Infinity Series Method Development Solutions support
automated method development workflows
• Method transfers from Agilent 1290 Infinity Series UHPC systems to other
(U)HPLC systems are seamless by ISET (Intelligent System Emulation
Technology)
• 1290 Infinity Series Method Development Solutions in combination with
ISET and additional QbD software are supporting fast method development
processes and method transfers to other (U)HPLC systems under QbD
aspects
• Agilent Remote Advisor and ACE are excellent solutions to support method
development workflows under QbD aspects for regulated environments
4
Introduction
• QbD (Quality by Design) is defined in the ICH guideline Q8(R2) http://www.fda.gov/downloads/Drugs/Guidances/ucm073507.pdf
• The ICH guidelines suggest to apply Quality by Design
principles in each step to eliminate risk or failures in drug
development processes
• Analytical method development for a drug is also a process
therefore quality principles from the ICH guideline should be
implemented
(ICH = International Conference of Harmonization of Technical Requirements
for Registration of Pharmaceuticals for Human Use, Founded in 1990 by an FDA initiative)
Traditional Development and Transfer of MethodsA chromatographic challenge !
5
± 10%
± 10oC
± 50%
± 10%
± 0.2
Slope 14.5% min MeOHpH 735oC1.0 mL/min
Fixed method specificationsdetermined manually
Transfer
Transfer
Develop ValidateTransfer
R & DQA / QC
6
Variability in resolution of criticalpeaks from day to day
Different results on different systems
Buffer solutions prepared bydifferent operators deliverdifferent results
Even small changes in pH show a large effect
Traditional Development and Transfer of MethodsReported Issues After Method Transfer
FDA Warning Letter Search for “Stability”Frequently stability issues reported
71 % of HPLC warning letters
Include Reference to “stability”
0 10 20 30 40 50 60 70 80
Calibration
Training
OOS
SSC
Method Validation
Defecient Methods
Data Integrity - Security
Data Integrity - Data Deletion
Data Integrity - Incomplete Records
Reaccurance
External Consultant Required
Data Review
Complete Data
Poor Response / CAPA
Stability
41
27
51
27
43
17
27
13
43
30
16
27
43
43
71
8
QbD Approach for Method Development
• Analytical QbD begins by defining goals (Analytical Target Profile, ATP)
and identifying potential method variables and responses
that affect method quality
• Statistical „Design of Experiments“ (DOE) has been applied to the
selected method variables leading to process and method
understanding and to create a list of critical method parameters
(CMPs- flow rate, temperature etc.) and critical method attributes
(CMAs- resolution, peak tailing)
• The experimentally measured responses were modeled to
determine the Design Space (defined in ICH Q8 (R2))
• A verified and validated method can be modified within the
Design Space to compensate any unforseen variables yet delivering
consistent results
± 10%
± 10oC
± 50%
± 10%
± 0.2
Slope 14.5% MeOH20mM bufferpH 745oC1.0 mL/min
Method specifications:
Traditional Approach• Fixed Protocol
Design space
QbD Approach• Variable Protocol
Traditional Approach vs. QbD Approach
9
Working within the design space will ensure the method‘s robustness
Content
• Introduction- Traditional Development and Transfer of Methods
- QbD Approach for Method Development
• Agilent Solutions for QbD Method Development-Agilent Method Development Systems
-Intelligent System Emulation Technology (ISET)
• QbD Method Development Workflow- Screening
- Optimization
- Robustness study, Design of Experiments
- Transfer & Verification
Agilent Solutions for QbD Method Development
ISET
Intelligent System Emulation Technology
RA
Remote Advisor
Harmonized Qualification
ACE
Method Development
System & Method Scouting SW
Agilent Method Development Systems
12
• 1352 different combinations of column chemistries and eluents
• A nearly infinite number of separation conditions is created by including
different temperature and flow rates as variable parameters
1290 Infinity II Method Development SolutionNew: Multi Column Thermostat
13
1290 Infnity I System 1290 Infnity II System
Autosampler
pump
detector
Multi Column Thermostat
MCT
1290 Infinity II Method Development Solution
Why different separation conditions?
P
a
g
e
1
4
min0.5 1 1.5 2 2.5 3 3.5
mAU
0
10
20
Eclipse plus C18
min0.5 1 1.5 2 2.5 3 3.5
mAU
0
10
20
Eclipse plus PhenylHexyl
min0.5 1 1.5 2 2.5 3 3.5
mAU
0
10
20
Bonus RP
Changes of a have a much higher effect than changes of N !
1290 Infinity II Method Development Solution
Aspects of Automated Method Development
15
• maximum method performance
• best combination of selectivity, resolution and speed
Analytical Efficiency
• maximum instrument utilization
• simple setup for method development
• no need for manual system changes
Instrument Efficiency
• maximum automation of routine tasks
• unattended data acquisition
• easy data evaluation
Laboratory Efficiency
link to video
1290 Infinity II Method Development Solution16
8
x
• 2x external solvent selection valves for up to 24 solvents plus built-in solvent valve for
another 2 solvents = 26 solvents in total
• 1x unique Agilent 8pos/18 port valve for up to 8 columns in a single MCT
• 8x individual heat exchanger in two temperature zones
Up to 1300 bar!
12 + 1
x
12 + 1
= 169
= 1352
1290 Infinity II
MCT
1290 Infinity II
Flexible Pump
External Solvent
Selection Valves
Up to 12 solvents
per SSV!!!
A2A1 B2B1
SSV2SSV1
Have access to > 1000 sets of unique separation conditions!
1290 Infinity II Method Development Solution
Product and Concept
17
http://www.chem.agilent.com/Library/brochures/5990-6226EN.pdf
Autosampler
Pump
Detector
New 1290 Infinity II LC
All in a single stack!
Up to 1300 bar!
1290 Infinity II Method Development Solutionproduct and concept
J
a
n
u
a
r
y
2
0
1
5
18
18
Infinity II
configuration
configuration
Infinity II modules
Valves… ….located in
comment
1 x 8pos/18port valve(G4239C)
1 x 1290 Inf II MCT (G7116B)
for access to up to 8
columns (column selection)
1 x 12pos/13port valve(G4235A)
1 x 1290 ext valve drive(G1170A)
1 x 12pos/13port valve(G4235A)
1 x 1290 ext valve drive(G1170A)
for access to up to 26
solvents (24 + 2 from pump)
Autosampler
Pump
Detector
Up to 1300 bar!
add column to column data base
1290 Infinity II Method Development Solutionproduct and concept – column centric view
J
a
n
u
a
r
y
2
0
1
5
19
19
one click column selection
20
1290 Infinity II Method Development Solution
product and concept – column centric view
• selected column is shown directly in the instrument control
• no need to deal with individual valve positions
January 2015
22
Agilent ChemStation Method Scouting Wizard
new features
p.81 ff, http://www.chem.agilent.com/Library/usermanuals/Public/G4230-90004_MDS_USR_EN.pdf
Feature Description Description
Version MSW A.02.03 / A.02.04 MSW A.02.05
Supported screening pumps • 1290 Infinity Binary Pump (G4220A)
• 1290 Infinity Binary Pump VL (G4220B)
• 1290 Infinity Quaternary Pump(G4204A)
• 1260 Infinity Quaternary Pump (G1311B)
• 1260 Infinity Binary Pump (G1312B)
• 1260 Infinity Bio-inert Quaternary Pump (G5611A)
• 1260 Infinity SFC Binary Pump (G4302A)
• 1290 Infinity Binary Pump (G4220A)
• 1290 Infinity Binary Pump VL (G4220B)
• 1290 Infinity Quaternary Pump(G4204A)
• 1260 Infinity Quaternary Pump (G1311B)
• 1260 Infinity Binary Pump (G1312B)
• 1260 Infinity Bio-inert Quaternary Pump (G5611A)
• 1260 Infinity SFC Binary Pump (G4302A)
• 1290 Infinity II High Speed Pump (G7120A)
• 1290 Infinity II Flexible Pump (G7104)
Support of pump clusters • Pump Valve Cluster with all pumps above
• 1260 Infinity Preparative Pump Cluster
(2x G1361A)
• Pump Valve Cluster with all pumps above
• 1260 Infinity Preparative Pump Cluster
(2x G1361A)
Support of multiple screening
pumps, as e.g. in SFC hybrid
system
If more than one pump is configured, as e.g. with the G4302A
plus LC pump, the user can select which pump will be used in
the screening campaign
If more than one pump is configured, as e.g. with the G4302A
plus LC pump, the user can select which pump will be used in
the screening campaign
Pump specific configuration
parameter
The default parameter used for flushing, equilibration, and
column storage can be stored separately for each pump
The default parameter used for flushing, equilibration, and
column storage can be stored separately for each pump
Supported column
screening configuration• Column compartment cluster of 2 or 3
1290 Infinity Thermostatted Column compartment
(G1316C)
• Column compartment cluster of 2 or 3
1290 Infinity Thermostatted Column compartment
(G1316C)
• Single 1290 Infinity II Multicolumn Thermostat (G7116B)*
with 6/14 or 8/18 valve
Report templates – Intelligent
Reporting
Several report templates are provided to support identification of
most interesting chromatograms
Several report templates are provided to support identification of
most interesting chromatograms
Automatic scheduling of
scouting sequences
The system will automatically schedule a
screening sequence (OpenLAB ChemStation)
The system will automatically schedule a
screening sequence (OpenLAB ChemStation)
*cluster of 2 or more MCTs is planned for later this year
January 2015
Setup Columns in TCC
(Edit Columns – Online Session Only)
Page 23
Column Setup in TCC
Page 24
Colors Corresponds to
Column Clips and
Tubing Collars
Setup Multi-Channel Valve Solvents
Page 25
Solvent Choices Appear in Pump Setup
Page 26
Method Development Scouting Wizard
Page 27
Choose Root Folder for Campaign
Description May Be Edited
Page 28
Campaign Uses Base Method From Chemstation
Define Screening Campaign Base:
For a method scouting campaign, a base method needs to be set up (you do this as
usual in the ChemStation
It is recommended to use Edit Entire Method
All parameters that are not changed are taken from this base method
These are typically the detector settings, autosampler settings, pump settings that
are not altered, and data processing parameters
Page 29
Choose Base Method & Screening Parameters
Page 30
Choose Columns to Be Used
Note: Column Properties Obtained From Column Table
Page 31
Choose Solvents to be Used
Note: Solvents Choices Obtained from Valve Table
Page 32
Setup Gradient(s) Profile(s)
Page 33
Gradients May be Setup with Timetable or
Graphically
Page 34
Analysis Steps: Optimized for Minimal
Number of Steps
Page 35
Column Temperature Screening
Page 41
Review and Select Methods
Note Methods are Flagged if Outside
Limits (Temp & pH)
Page 42
Exclusion may be overridden
43
Method Scouting Wizard Reporting
All available chromatographic information can be used for
reporting - across-samples, across-sequences, across-labs*!
Tables, diagramms, conditional formating.... are available.
*) across labs - only in combination with OpenLAB ECM
44
Method Scouting Wizard Reporting
Scouting
conditions
sorted for
highest
number of
peaks
Many options for sorting and
filtering....
45
Method Scouting Wizard Reporting
Experiments vs.
Maximum retention
time and maximum
number of peaks
(bubble-size)
Experiments vs.
retention time of
peaks (projection of
all chromatograms)
Many options for graphical overviews,
e.g. ...
46
Method Scouting Wizard Reporting
Data-file, column,
solvents, temperature,
gradient-name from
Method Scouting Wizard
And of course chromatograms – with scouting
conditions!
Intelligent System Emulation Technology (ISET)
47
- Seamless transfer of methods between LCs, regardless of the brand
PO-Nr. G2197AA
Method Transfer Between Different LC Instruments
48
min3 4 5 6 7 8 9 10 11
mAU
0
25
50
75
100
125
150
175
1100 Series Binary System
1290 Infinity LC
Method transfer from a UHPLC system with a minimized dwell
volume and optimized mixing behavior to any other LC system is
often challenging and affects rentention time and resolution
Example: Method Transfer from a 1290 UHPLC to a 1100 HPLC system
49
1290 Infinity LC
min0 1 2 3 4 5 6 7 8 9
mAU
0
100
200
300
400
500
600
1100 Series Quaternary LC
Gradient differences due to different dwell
volumes and different mixing behavior
Method Transfer Between Different LC Instruments
Results from a tracer experiment
min0 1 2 3 4 5 6 7 8 9
mAU
0
100
200
300
400
1290 Infinity LC
1 min isocratic hold
1100 Series Quaternary LC
50
Method Transfer Between Different LC InstrumentsApproach # 1: Isocratic holding step to synchronize
Results from a tracer experiment
Still gradient differences due to
different mixing behavior
1260 Infinity LC
1290 Infinity LC
+ 900 µl hold
51
Approach # 1: Applying Isocratic Holding Steps
Results
• Results shows a low consitency
• Requires manual determination of the dwell volume/ isocratic hold
(in solvent delivery systems equipped with dampeners the dwell volume is pressure
dependent and variable)
• Requires modification of the methods
(should be avoided in validated environment,
but doesn’t require revalidation USP Chapter <621>)
min0 1 2 3 4 5 6 7 8 9
mAU
0
100
200
300
400
1290 Infinity LC
1 mL loop installed
1100 Series Quaternary LC
52
Method Transfer Between Different LC InstrumentsApproach # 2: Adding a physical void volume
Almost consistency of both
gradient curves
Results from a tracer experiment
1290 + 1000 μL dwell vol.
1100/1200 Quat Pump
Approach # 2: Adding a Physical Void Volume
Results
53
• Results shows a good consitency
• Manual determination of dwell volumes required (issues of a variable
dwell volume with systems containing damperners)
• All mechanical changes are laboriously not flexible
Agilent Solution:
Intelligent System Emulation Technology (ISET)
54
Programmed gradient
1290 gradient
1200 gradient
0 0.5 1 1.5 2 2.5 3 3.5
0
50
100
150
200
250
300
350
400
min
mAU
InjectionSoftware controlled compensation
of dwell volume differences and
synchronization of mixing
behaviors
Agilent Solution: Method Transfer by ISETResults: 1260 Infinity Binary LC to 1290 Infinity LC
55
min1 2 3 4 5 6 7 8 9
mAU
0
10
20
30
40
Imp
A
Imp
B
Imp
H Imp
J
Imp
F
Imp
K
Pa
race
tam
ol
1290 Infinity LC with ISET
1260 Infinity Binary LC
• Consistency of results
1290 Infinity Quaternary Pump - BlendAssist
Simple tool for online-dilution of modifiers and gradient set-up
You need different concentrations of modifiers in your analysis, would like to have just one stock-
solution and do online dilution to profit from the quaternary mixing capability of your pump? Here is a
simple tool – BlendAssist!
Water Water
1% TFA
ACN
ACN
1% TFA
Desired method conditions - example:
1. 5 to 95% gradient of ACN with 0.1% TFA in
Water and 0.08% TFA in ACN
2. 20 – 80% gradient of ACN with 0.5% TFA in
Water and 0.4% TFA in ACN
Without BlendAssist you need to either pre-mix the
required solvents or by using stock-solutions of
TFA in Water and ACN to program complex
gradients (%A, B, C, D).
With BlendAssist: just program your binary
organic/aqueous gradient and define the dilution
factor!
56
1290 Infinity Quaternary Pump - BlendAssist
Simple tool for online-dilution of modifiers and gradient set-up
57
min1 2 3 4
Norm.
0
5
10
15
20
25
30
35
40
6 dynamically mixed mobile phases
6 premixed mobile phases (same user)
min1 2 3 4
Norm.
0
5
10
15
20
25
30
Analysis of Glucocorticoids - Comparing dynamically mixed mobile phase vs. user influence
58
A:B=70:30
A:B=70:30
Buffer Advisor SoftwareQuaternary mixing
February 3, 201759
Buffer Advisor SoftwareWorkflow
67
High Dynamic Range (HDR)Increasing the linear range by a factor of >30
1260/1290 Infinity DAD
> 30 x Linearity
with HDR
1.4 1.8 2.01.3
>10
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
1040 > 1090 1090 > 1050 1050 > 1100 1100 > 1200 1200 >1290
History of DAD Sensitivity Gain - The last 30 years
Detector Model (Intro Year)
Sensitivity
Gain
(1988)(1986) (1995) (2006) (2010)
1260 / 1290 Infinity
Diode Array Detector
Page 68
History of DAD Linearity Gain - The last 30 years
Detector Model (Intro Year)
(1988)(1986) (1995) (2006) (2010)
Page 69
1.4 1.8 2.0 1.42.9
39.1
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
1040 to 1090 1090 to 1050 1050 to 1100 1110 to 1200 1200 to 1290/60 1200 to HDR
Linearity
Gain
(2011)
1260 / 1290 Infinity
Diode Array DetectorWith HDR
>30X Linearity
Deuterium
Lamp
Max-Light Cartridge Cell
10 mm or 60 mm path length
Programmable
or fixed slit
1024 element
diode-array
GratingMirror
Optofluidic Waveguides: Max-Light Flow Cells - Total-internal reflection in a non-coated fused silica fiber
1260 / 1290 Infinity DAD
Page 70
Effects of path length increase
Peak Dispersion
60 mm pathlength
large geom. cell volume (78 µL)
lower light transmission
- Loss of resolution
- Loss of signal height
10 mm pathlength
small geom. cell volume (13 µL)
high light transmission
Conventional flow cells:
Page 71
60 mm pathlength
4 µL σV dispersion volume
Max-Light High Sensitivity cell:
Optofluidic waveguides
(total internal reflection)
10x Higher Sensitivity 1290/1260 Infinity DAD compare to 1200 Series DAD SL
1260/1290 DAD
60 mm
1200 DAD SL
10mm
Height [mAU] 28.876 4.938
Noise [mAU] 0.0098 0.0190
Signal/ noise 2944 259
Increase 11.4 x
min0 1 2 3 4 5
mAU
0
5
10
15
20
25
30
Columns: 150 x 4.6mm Zorbax SB C18, 5µm
Sample: Anthracene: 835 pg/µL
Mobile phase: A: Water, B: Acetonitrile
Elution: isocratic 80 % B
Injection volume: 5 µL
Flow: 1.5 mL/min
DAD: 251/4nm, Ref= 450/80nm,
2.5Hz, slit width 4nm
1200 Series
DAD / DAD SL
1260/1290
Infinity DAD
Page 72
Max-Light cartidge cells for 1260/1290 DAD- Specifications
73
10 mm path length 60 mm path length
Max-Light
cartidge cell
Path
length
Dispersion
volume σv
Noise Noise
per cm
Standard 10 mm 1.0 μL ± 3.0 µAU ± 3.0 µAU/cm
High Sensitivity 60 mm 4.0 μL ± 3.6 µAU ± 0.6 µAU/cm
Linearity, linear range and dynamic range- 1200 Series Diode Array Detector, 10 mm flow cell
Page 74
Conc [µg/mL]
Detector Signal
[AU]
LOD
(3x S/N)Conc (upper limit)
linear range
Detector
noise
7*10-6
Conc (upper limit)
dynamic range
Linear range
Dynamic range
2.0
Slope = Response Factor
(Sensitivity)
1260/1290
DAD
1260/1290
DAD
Upper limit
5
10_3
__
NoiseDetector
LimitUpperrangeLinear
10 mm
Linearity, linear range and dynamic range- 1260/1290 Infinity Diode Array Detector, 10 mm flow cell
Page 75
Conc [µg/mL]
Detector Signal
[AU]
LOD
(3x S/N)Conc (upper limit)
linear range
Detector
noise
3*10-6
Conc (upper limit)
dynamic range
Linear range
Dynamic range
2.5
typically
~ 3 times higher linear range vs. 1200 Series DAD
(due to lower detector noise, higher upper limit)
10 mm
Linearity, linear range and dynamic range- 1260/1290 Infinity Diode Array Detector, 60 mm flow cell
Page 76
Conc [µg/mL]
Detector Signal
[AU]
LOD
(3x S/N)
6 times higher signal
Conc (upper limit)
linear range
Detector
noise
3.6*10-6
Conc (upper limit)
dynamic range
Linear rangeDynamic range
2.5
typically
Shift of linear and dynamic range to lower concentration,
but same linear range: Gain at lower concentration,
loss at higher concentration)
Combine two flow cell
with short and long path length
60 mm
Linearity, linear range and dynamic range- 1260/1290 Infinity DAD with HDR (3.7 mm and 60 mm flow cell)
Page 77
Conc [µg/mL]
Detector Signal
[AU/cm]
LOD
(3x S/N)Conc (upper limit)
linear range
Detector
noise
0.6*10-6
Conc (upper limit)
dynamic range
Linear range
Dynamic range
6 .7
60 mm
Flow cell 3.7 mm
Flow cell
+One normalized signal AU/cm resulting in linear range
> 3.7 x 106 (39x higher than 1200 Series DAD)60 mm 3.7 mm
78
3.7mm
60mm
Clustered DAD
1290/1260 High Dynamic Range – HDR Increasing Linear UV Range by >30x
Technology:
Cluster of two DADs
• DAD 1:
3.7mm Cell – for high concentrations
• DAD 2:
60 mm Cell – for low concentrations
• Output: One, combined signal,
normalized to 10 mm path length
(HDR range: 0.6 x 10-6 to 6.7 AU/cm)
vs for 1200 Series: 7 x 10-6 to 2 AU/cm)
Control / Usability
• Like standard 1290/1260 DAD:
Control, data analysis, reporting
Investment Protection
• Existing 1290/1260 DAD can be upgraded
(by 2nd DAD)
• Existing 1100/1200 Systems can be
upgraded (by two 1290 or 1260 DAD)
DAD 1:
DAD 2:
High Dynamic Range (HDR) Signal
Page 79
The combined signal is
dominated by the long path
The combined signal is dominated by the short path
Transition range
Impurity Analysis, DAD with 10 mm flow cell
February 3, 2017Confidentiality
Label
81
min2 3 4 5 6 7
mAU
0
2
4
6
8
0.07 %
S/N 21
0.63 % 0.02 %
S/N 3.2
0.68 %
0.01 %
S/N 1.4
3.83 %
0.002 %
S/N 0.4
9.99 %
0.05 %
S/N 14
0.70 %
min2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8
mAU
0
500
1000
1500
2000 Impurity level
Signal/Noise
Area rsd %
Impurity Analysis, DAD with 60 mm flow cell
February 3, 2017Confidentiality
Label
82
min2 3 4 5 6 7
mAU
0
2
4
6
8 0.07 %
S/N 52
0.38 % 0.02 %
S/N 7.9
0.29 %
0.01 %
S/N 3.6
0.90 %
0.002 %
S/N 0.9
7.92 %
0.05 %
S/N 35
0.35 %
min2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8
mAU
0
500
1000
1500
2000
2500
3000
3500
Impurity level
Signal/Noise
Area rsd %
Improved S/N and RSD
Main not within linear range
Impurity Analysis, DAD HDR
February 3, 2017Confidentiality
Label
83
min2 3 4 5 6 7
mAU
0
2
4
6
8
0.07 %
S/N 53
0.35 % 0.02 %
S/N 8.0
0.30 %
0.01 %
S/N 3.6
1.57 %
0.002 %
S/N 0.9
11.1 %
0.05 %
S/N 35
0.35 %
min2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8
mAU
0
250
500
750
1000
1250
1500
1750
2000
Impurity level
Signal/Noise
Area rsd %
Improved S/N and RSD
Main within linear range
Impurity Analysis, DAD HDR, 3x Inj. Vol. (3µL)
February 3, 2017Confidentiality
Label
84
min2 3 4 5 6 7
mAU
0
2
4
6
8
0.07 %
S/N 159
0.30 % 0.02 %
S/N 29
0.28 %
0.01 %
S/N 13
0.48 %
0.002 %
S/N 3.2
4.88 %
0.05 %
S/N 108
0.28 %
min2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8
mAU
0
500
1000
1500
2000
2500
3000
3500
Impurity level
Signal/Noise
Area rsd %
Further Improved S/N and RSD
Main still within linear range
87
1290/1260 High Dynamic Range – HDR Increasing linear UV Range by 30x
AdvantagesImpurity Analysis
• Main + impurities in one run
• 30x lower LOD/LOQ (vs 1200)
• 15x lower LOD/LOQ (vs 1290/1260)
• 3 x higher upper limit of linearity
• ID / Quant of lower level
impurities (0.002% / 0.01%)
• Ideal for fixed dose combinations
• Prepared for stronger regulations
• Genotoxins
Improved Area Precision enabling
Automated Integration
• Higher Productivity
• Less error-prone
• Higher data quality and security
3.7 mm
60mm
Clustered DAD
ChromSword
AutoChrom (ACD Labs)
Fusion QbD (S-Matrix)
Add-on Software Options for QbD Method Development
4.5
Rs = 9.3 + 4.2(∆T) - 5.4(%I)2 + 12.7(pH)2 + 1.3(∆T *pH) + 1.6(∆T)2%I+ …
Linear Effects Curvature
Effects
Complex EffectsInteraction
Effects
pH
Oven Temp
Initial %
Organic
5.0
15.0
30.0 60.0
90
2.5
3.5
Fusion QbD Software (S-Matrix)
• Uses a DOE and statistical modeling approach.
• Example equation below illustrates resolution is a complex function of different
interacting parameters (temp, flow, pH, etc.).
February 3, 201791
Details of Fusion QbD Software (S-Matrix)Variable ConstantColumn (Agilent ZORBAX RRHD Eclipse Plus
Phenyl-Hexyl) length
• Gradient
Equilibration : 1.0 min
5%B
Initial hold: 1 min, 5% B
Final Hold: 2 min at
95%B
Re-equilibration: 4min
at 5%B
• Flow Rate: 0.6 mL/min
• Injection volume: 1 µL
• Wavelength: 245 nm ±
4 nm (ref off)
• All other finalized
parameters from
Phase 1
• 3.0x50 mm, 1.8 µm (p/n: 959757-312)
• 3.0x100 mm, 1.8 µm (p/n: 959964-312)
Solvents
• A1: pH 4.0, 5 mM formic acid and 10 mM
ammonium formate in water
• A2: pH 4.5, adjusted from pH 5 with
acetic acid
• A3: pH 5.0, 5 mM acetic acid and 10 mM
ammonium acetate in water
• A4: pH 5.5, adjusted from pH:7 with
acetic acid
• A5: pH 6.0, adjusted from pH:7 with
acetic acid
• A6: pH 6.5, adjusted from pH:7with acetic
acid
• A7: pH 7.0, 10 mM ammonium acetate in
water
• B1: Acetonitrile
• B2: Acetonitrile: THF (88:12)
Gradient time
• 9 min
• 15 min
Column temperature
• 35°C
• 40°C
• 45°C
• 50°C
Response Goals Target Lower
Bound
Upper
Bound
Relative
Rank
No. of Peaks Maximize 36 45 1
No. of Peaks >= 1.50 (Tangent
Resolution)
Maximize 28 34 0.9
No. of Peaks >= 2.00 (Tangent
Resolution)
Maximize 21 28 1
Max Peak #1 (Tangent Resolution) 2.51 1.40 3.62 1
Max Peak #1 (Symmetry) Minimize 1.14 1.15 0.9
left side:
Example of Parameters used to optimize
methods
bottom:
Example: Trend Response goals used in
the chemistry screening phase
List of CMPs and CMAs for robustness
testing
* The coded names are used in robustness model displays
Details of Fusion QbD Software (S-Matrix)Robustness Testing & Establishing Design Space (MODR)
Design Space
93
31
Target System
Emulation by ISET
Robustness study by
QbD software
1290
R&D
1260QA/QC
Method transfer
Method development
System
Use of 1.8 µ particles
and QbD software
Target Systems in
QA/QC labs
QbD Method Development & Method Transfer WorkflowFrom UHPLC to HPLC in a nutshell
2
1290
R&D
1260
R&DEmulation
94
Overall workflow which
consists of four main steps
namely
• Step # 1: Screening
Screening
• Column Chemistry
• pH conditions
• Organic Solvent Selection
• Gradients, temperatures
Agilent Method
Scouting
Wizard
QbD Based Method Development Workflow
95
Step # 1: ScreeningResults
For more details please see Agilent Application Note 5991-0989EN
Bubble size represents the number of integrated peaks and, consequently, best
mobile and stationary phase combination
96
Overall workflow which
consists of four main steps
namely
• Step # 1: Screening
• Step # 2: Optimization
Screening
• Column Chemistry
• pH conditions
• Organic Solvent Selection
• Gradients, temperatures
Agilent Method
Scouting
WizardOptimization
• Optimize Gradient Profiles
• pH conditions
• Flow rates, temperaturesQbD Software
QbD Based Method Development Workflow
97
Variable Parameters Study Range
Pump Flow rate (mL/min)
Intermediate hold time (min)
Final % Strong Solvent (Gradient 1)*
Oven Temperature (°C)
0.550, 0.600,0.650
3.00 <= Intermediate Hold Time <= 7.00
30.0 <= Final % Strong Solvent <= 35.0
33.0, 36.0, 39.0
Constant Parameters Constant Value
Column Type
Wavelength
Strong Solvent type
pH
Injection Volume
Equilibration Time
Initial Hold Time
Gradient 1 Time*
Gradient 2 Time*
Final Hold Time
Final Hold % Organic
Initial % Strong Solvent
3.0X100 mm,1.8 µm ZORBAX RRHD Eclipse Plus Phenyl-Hexyl
245 nm ± 4 nm (ref off)
Acetonitrile
7.0
1µl
2.50 min
1.00 min
5.17 min
9.28 min
2.00 min
90.0 %B
5.0% B
Step # 2: OptimizationFusion AE QbD Software
Gradient 1: 5%B-(30-35)%B
Gradient 2: (30-35)%B-90%B
Optimization of flow rate, gradient slope, pH, column temperature
99
Overall workflow which
consists of four main steps
namely
• Step #1: Screening
• Step #2: Optimization
• Step #3: Robustness study
Screening
• Column Chemistry
• pH conditions
• Organic Solvent Selection
• Gradients, temperatures
Agilent Method
Scouting
WizardOptimization
• Optimize Gradient Profiles
• pH conditions
• Flow rates, temperaturesQbD Software
QbD Software
Design of Experiments
• Multivariate study
• Robust region
• Design Space
QbD Based Method Development Workflow
3D Graph Called
Response
Surface
Example - Why QbD is Needed…..Step # 3: Design of ExperimentsInvestigate multivariate relationships
Center Point of
Robust region –
Prediction
Center Point of
Robust region –
Experiment
API-Symmetry 0.71±0.04 0.66
API-tailing USP 1.4±0.07 1.4
API-Tangent
Width
0.081±0.004 0.083
API-Resolution 2.3±0.3 2.0
Optimized Parameters
Moving critical pair in
a more robust region
101
Step #3 : Design of Experiments
Results: Design SpaceCritical Method
Parameters (CMPs)
Proven
Acceptable
Range (PARs)
Critical Method
Attributes (CMAs)
Column: Agilent ZORBAX
RRHD Eclipse Plus C8 3.0X50
mm, 1.8 µm
No. of peaks (>40)
API resolution (>1.5)
Peak purity (≥ 98%)
Peak tailing (<1.5)
Strong solvent: Methanol
% Strong solvent: 90.5% ± 1.5%
Aqueous solvent pH: 7.7 ± 0.1
Gradient range: 5% to 90.5%
Oven Temperature: 45°C
Gradient time: 15 min
Flow rate: 0.6 mL/min
Wavelength: 292 nm
CMAs to create a Design Space
The Design Space is a region in which changes to method parameters will not
significantly affect the results.
102
Overall workflow which
consists of four main steps
namely
• Step #1: Screening
• Step #2: Optimization
• Step #3: Robustness study
• Step #4: Method Transfer
& Verification
Screening
• Column Chemistry
• pH conditions
• Organic Solvent Selection
• Gradients, temperatures
Agilent Method
Scouting
WizardOptimization
• Optimize Gradient Profiles
• pH conditions
• Flow rates, temperaturesQbD Software
QbD Software
ISET
QbD Software
Design of Experiments
• Multivariate study
• Robust region
• Design Space
Method Transfer &
Verification
• Agilent Method Translator
• ISET
• QbD Optimization
• Verification
QbD Based Method Development Workflow
103
3
Target System
Emulation by ISET
Robustness study by
QbD software
1260QA/QC
Method transfer
Target Systems in
QA/QC lab
Verification
Step # 4: Method Transfer & VerificationFrom UHPLC to HPLC
2
1290
R&D
1260
R&DEmulation
104
HPLC design space with new CMA
Robustness Study After Transfer Modeling a HPLC design space on an emulated 1260 system
Critical Method
Parameters (CMPs)
Proven
Acceptable
Range (PARs)
Critical Method
Attributes (CMAs)
Column: Agilent ZORBAX
Eclipse Plus C8 4.6X150
mm, 3.5 µm
No. of peaks (>40)
API resolution (>4)
Peak purity (≥ 98%)
Peak tailing (<1.3)
Strong solvent: Methanol
% Strong solvent: 87.5.% ± 1.5%
Aqueous solvent pH: 7.7 ± 0.1
Gradient range: 5% to
87.5%
Oven Temperature: 37°C
Gradient time: 45 min
Flow rate: 1.4 mL/min
Wavelength: 292 nm
HPLC design space parameters
105
Proof Of Robustness After TransferConditions applied from center point and the four corner points of the Design Space
min10 20 30 40 50
mAU
050
150
250
350
min10 20 30 40 50
mAU
050
150
250
350
min10 20 30 40 50
mAU
050
150
250
350
min10 20 30 40 50
mAU
050
150
250
350
min10 20 30 40 50
mAU
050
150
250
350
% B max: 86 %; pH: 7.6
% B max: 89 %; pH: 7.6
% B max: 87.5 %; pH: 7.7
% B max: 89 %; pH: 7.8
% B max: 86 %; pH: 7.8
A
B
D
T
C
API Rs> 4 and API tailing = 1.2 for all runs
Critical Method Attributes remain within the limits
Verification Of The Final Method With The Target SystemResults: 1260 Infinity data compared to 1290 Infinity data in emulation mode
106
Final gradient
Time %B
0.3 5
45.6 87.5
49.2 87.5
49.5 95
49.8 95
50.1 5
53.1 5
mAU
24.0
91
API Rs = 4.2
API tailing = 1.2
min10 20 30 40 50
0
100
200
300
400
23.3
44
1290 emulated as 1260
mAU
100
200
300
400
min10 20 30 40 50
API Rs = 4.1
API tailing = 1.2
0
23.3
47
24.1
15
1260 target system
Critical Method Attributes are within the defined limits
Agilent Application Note
February 3, 2017107
Title: QbD Based Method Development on an
Agilent 1290 Infinity UHPLC system Combined
with a Seamless Method Transfer to HPLC
Using Intelligent System Emulation Technology
Type: Application Note
Publication Number: 5991-5701EN
Pages: 8
Target segments: Pharmaceutical QA/QC
Language: English
Author: Vinayak A.K.
LitStation Availability: May 5, 2015, CPOD
108
*An extra HPLC optimization time is added after the method transfer to conventional particle sizes to optimize CMAs.
Experiments Sub-2-µm columns
(time in hours)
Conventional columns
(time in hours)
Screening 20 47
Optimization* 28* 20
Total 48 67
Conclusion
• Due to shorter runtimes when using sub-2-micron columns the efficiency
of the method development process has been more than doubled
• ISET can be used to emulate seamless different target systems
• QbD software will deliver robust methods where the critical method
attributes remain constant
• No need for revalidation of methods when applying CMP which are
within the design space
Overall Summary
• QbD is a state of the art approach to remove variability of
analytical methods
• ISET enables cross platform method development workflows
• Method Scouting Wizard Software enables easily to create
sequence tables for method development workflows
• The combination of Agilent 1290 systems, ISET, Method
Scouting Wizard Software, Blend Assist, Buffer Advisor, and
third party QbD software provides a unique and efficient way
to develop and transfer robust methods
109
February 3, 2017
Confidentiality Label
110
THANK YOU
QbD process terminology Analytical QbD
terminology
Examples
Analytical Target Profile
(ATP)
Accurate quantitation of API
without interferences from
degradants
Quality Target Product
Profile (QTPP)
Quality Target Method Profile
(QTMP)
pKa, Log P, Solubility
Critical Process Parameters
(CPP)
Critical Method Parameters
(CMP)
Flow rate,
Temperature, pH
Critical Quality Attributes
(CQA)
Critical Method Attributes
(CMA)
Resolution, Peak
Tailing, Peak Capacity
Control Strategy pH ± 0.1; Wavelength
± 2 nm
QbD terminology in Method DevelopmentAppendix
Analytical QbD Terminology
Appendix : Agilent Application Notes
QbD based Method Development
• Quality-by-Design Approach to Stability Indicating Method
Development for Linagliptin Drug Product
– Agilent Application Note 5991-3834EN
• Automated QbD Based Method Development and Validation of
Oxidative Degraded Atorvastatin
– Agilent Application Note 5991-4944EN
• Development of an UHPLC Method for Azithromycin Tablets Using
ChromSword Auto Software
– Agilent Application Note 5991-5428EN
• QbD Based Method Development on an Agilent 1290 Infinity UHPLC
system Combined with a Seamless Method Transfer to HPLC Using
Intelligent System Emulation Technology
- Agilent Application Note 5991-5701EN
112