developing an analytical impurity control strategy using · pdf fileimpurities •...
TRANSCRIPT
Developing an Analytical Impurity Control Strategy Using QbD
Mark D. Argentine
Quality by Design in DevelopmentRelies upon developing knowledge around processes and products…
… for effective process and
product design and control.
Ref: M. Nasr, 2006
2IFPAC 2014M. Argentine
A + B C API
(D1, D2 …)
solvent 1 solvent 2, reagent
time, temp.,pH …(B’) (C’) (API’)
(BP1, BP2 …)
A, B = starting materialsC = intermediateAPI = active pharmaceutical ingredientB’ = starting material impurity with potential
to form C’ and API’BP = reaction by-productD = degradation product
An understanding of “What goes in” and “What goes on” in development…
3
…leads to knowledge and development of suitable, robust manufacturing and analytical controls for your process
IFPAC 2014M. Argentine
Building a Process Knowledge BaseConsider multiple synthetic variations, suppliers for starting materials, etc.
Utilize chemistry-guided and knowledge-guided approaches to potential impurities
• Knowledge-guided: What has been observed in selected samples• Chemistry-guided: What is possible/likely from and understanding of the synthetic
processes used or considered– Any potential genotoxic impurities likely/possible? Use “fit-for purpose” development methods to
explore for potential impurities.
Leverage appropriate tools to develop process and impurity formation/fate understanding - e.g., MS, NMR, IR/NIR/Raman
Control strategy development is a dynamic partnership between process and analytical and an iterative strategy
• To search for and identify impurities• To understand impurity formation and fate • To design appropriate process and analytical controls to minimize or eliminate impurities.
IFPAC 2014M. Argentine
4
Example: An HPLC-IR-NMR system to generate development knowledge
5
NMR
To NMR
From NMR to HPLC
Reactor
Pump
HPLC(configured for on-line)
From HPLC to Reactor
From NMR to HPLC
Optical Spectrometer(mid-IR, NIR, Raman, UV-vis)
Obtain valuable information on
kinetics, reactive
intermediates, relative
response factors, etc.
AU
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
0.045
0.050
Minutes0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00
Step 2
Step 3
Step 1
90%10%
90%80% 10%20%
70%30%
10%90%
Impurity Controls Based upon Process Understanding
>98%
NMT 0.50%
NMT 0.15%
NMT 0.40%
NMT 1.0%
NMT 0.25%NMT 0.15%
Imp B’’
Imp B’Imp A’
Imp A
A
B
Imp B
Imp C’
Imp C
C
D
Imp D’
Imp D
Imp D’’
NMT 0.15%
NMT 0.10%
NMT 0.10%
6IFPAC 2014M. Argentine
Process knowledge informs analytical controls
Knowledge space studies
Understanding of measurement requirements for the quality attributes of a product that must be controlled
• What analyte(s)• What level(s)• What precision
Analytical method design space studies
7
An Analytical Target Profile (ATP)
IFPAC 2014M. Argentine
Building an Analytical Knowledge Base -Control Strategy Development
Development MethodsHPLC broad polarity screens, multiple detectors
orthogonal techniques, on-line analysistargeted methods
Quality Control MethodsIntegral to specifications and/or process controls
Optimized for ruggedness
Process knowledge - what needs to be monitored/controlled (Analytical Target Profile definition)
8IFPAC 2014M. Argentine
QbD Tools in Impurities Method Development Strategy
• Identify key impurities from:– Typical drug substance/product samples
– Authentic impurity samples
– Process development samples with likely impurities (e.g., reaction samples, mother liquors)
– Samples from designed, forced degradation studies
– Potential drug product impurities --- review degradation and excipient interaction design study data
• Use systematic tools and designed studies to aid in appropriate method development (e.g., column screening program for design space impurities)
Design SpaceDesign Space
9IFPAC 2014M. Argentine
An example: Analytical Design Space for Impurity Method Development
Zorbax SB C8
Bonus RP
ACE Pheny l
Xterra MS C18
ACE CN
Atlantis dC18
Polaris C8-Ether
Alltima C18
Acquity UPLC BEH C18
Acquity UPLC BEH C8
Acquity UPLC BEH Pheny l
Acquity UPLC BEH Shield RP18
Acquity UPLC BEH HSS T3
SunFire C8Xbridge C8
Xbridge C18
Hypersil Gold PFP
Hypersil Gold aQ
Gemini C6-Pheny l
Luna C18(2)-HST
Halo C18Zorbax Eclipse Plus C18
Discov ery HS F5
Hypersil Gold C18
Microtech C18
Platinum C18
Platinum EPS C18
Sy nergy Max-RP
Sy nergy Hydro-RP
Gemini C18
Use column classification system*
to identify column similarities and
differences
And
Exploit column differences in HPLC method development
*e.g., Gilroy, Jonathan J.; Dolan, John W.; Snyder, Lloyd R. Column selectivity in reversed-phase liquid chromatography IV. Type-B alkyl-silica columns. Journal of Chromatography, A (2003),1000(1-2), 757-778.
10IFPAC 2014M. Argentine
AU
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
Minutes1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00
4
2 7
15
3
64
27
1
3
6
MethanolAcetonitrile
5
Column 1 - C8, low pH
Column 2 – Polar Embedded, low pH
AU
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
Minutes1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00
4
2
1
53 6,7
4
2
1
3
6,7
MethanolAcetonitrile
AU
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
Minutes1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00
4
2 7
15
3
64
27
1 5
3
6
MethanolAcetonitrile
Column 3 – Phenyl, low pH
AU
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
0.24
Minutes1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00 11.00 12.00 13.00
427 1
5
6
4
2
7
15
6
MethanolAcetonitrile
Column 4 – C18, high pH
33
5
AU
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
0.18
0.20
0.22
Minutes2.00 4.00 6.00 8.00 10.00 12.00 14.00
6
17
2
4
3
Perform screen with distinctly different conditions/phases
Optimize separation with modeling5
Predictions for directing additional studies
Assessing method “robustness” w/o doing experiments – power of modeling tools
12
evaluate parameters such as temperature, organic strength, and gradient sensitivity for impact on expected
minimum resolutionIFPAC 2014M. Argentine
mV
50.00
51.00
52.00
53.00
54.00
55.00
56.00
57.00
58.00
59.00
60.00
Minutes0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
1 2 34 5 6
7
A
B
Method selectivity of
gradient:
A= unspikedmatrix
B=matrix with 0.05%
spike for impurities
1,2,6,7,8 and impurity 5 spiked at
0.15%
13
8
Gradient Assay Identified in Development for Impurity Control
Historically, a gradient assay has been used in development to evaluate material quality and confirm the lack of late-eluting impurities
IFPAC 2014M. Argentine
110.00
105.00
100.00
95.000
90.000
85.000
80.000
75.000
70.000
65.000
60.0000 295 590 885 1180 1475 1770 2065 2360 2655 2950
TIME (SECONDS)
LOT G
LOT F
LOT E
LOT D
LOT C
LOT B
LOT A
BLANK
Initial Gradient Method – no significant later-eluting impurities for drug
14IFPAC 2014M. Argentine
85.000
82.700
80.400
78.100
75.800
73.500
71.200
68.900
66.600
64.300
62.0000 120 240 360 480 600 720 840 960 1080 1200
TIME (SECONDS)
SYS SUIT
LOT G
LOT E
LOT D
LOT C
LOT B
LOT A
BLANK
Impu
rity
1 (D
I)
Impu
rity
2 (D
I)
Can be simplified to an isocratic assay for routine control
Impu
rity
3 (D
I)Im
purit
y 4
(PI)
Impu
rity
5 (P
I)
Impu
rity
6 (D
I)
15IFPAC 2014M. Argentine
System Suitability Definitionm
V
18.00
20.00
22.00
24.00
26.00
28.00
30.00
32.00
34.00
36.00
38.00
Minutes0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00
Blank
System Suitability
Impu
rity
4
AP
I
Impu
rity
5
Impu
rity
6
CP 1 CP 2 Late eluting marker
16
Identify meaningful system suitability controls to ensure reliable method performanceIFPAC 2014M. Argentine
17
Example Chromatography Using an API Sample
73.000
72.000
71.000
70.000
69.000
68.000
67.000
66.000
65.000
64.000
63.0000 120 240 360 480 600 720 840 960 1080 1200
TIME (SECONDS)
SYSTEM SUITABILITYIm
purit
y 4
Impu
rity
5
Impu
rity
6
AP
I
API
Use of system suitability samples and a “method performance sample” (if it contains measurable impurity levels and is stable) can be useful for method robustness and method transfers
IFPAC 2014M. Argentine
18
Using Design Studies to Evaluate Method Robustness
Patten # DesignColumn Temp
(±3ºC)%ACN-Start
(±2%)%ACN-End
(±2%)%TFA
(±0.01%)Flow Rate
(±0.1mL/min)Iso. Hold (±0.5min)
pattern 0 0 40 18 80 0.1 1.5 5
pattern 1 ++++++ 43 20 82 0.11 1.6 5.5
pattern 2 ++−−−− 43 20 78 0.09 1.4 4.5
pattern 3 −+−+−+ 37 20 78 0.11 1.4 5.5
pattern 4 −−++−− 37 16 82 0.11 1.4 4.5
pattern 5 +−+−−+ 43 16 82 0.09 1.4 5.5
pattern 6 +−−++− 43 16 78 0.11 1.6 4.5
pattern 7 −−−−++ 37 16 78 0.09 1.6 5.5
pattern 8 −++−+− 37 20 82 0.09 1.6 4.5
pattern 0 0 40 18 80 0.1 1.5 5
Consider design studies during method development for more method knowledge
Leverage modeling studies (e.g., DryLab) to cover wide
operating ranges, when possible, in final method
IFPAC 2014M. Argentine
Robustness – Statistical Design Results
Patten # Design Standard Area Total Impurities (Direct area-%)
Total Impurities (vs ext std))
Rs (resolution pair 1)
Rs (resolution pair 2)
pattern 0 000000 10002 0.68 0.71 1.96 3.71
pattern 1 ++++++ 9908 0.69 0.67 1.88 2.38
pattern 2 ++−−−− 10614 0.69 0.71 2.00 3.43
pattern 3 −+−+−+ 10759 0.68 0.70 1.89 5.20
pattern 4 −−++−− 9985 0.69 0.77 1.92 5.34
pattern 5 +−+−−+ 10384 0.67 0.73 2.03 2.18
pattern 6 +−−++− 8978 0.67 0.74 2.09 3.41
pattern 7 −−−−++ 9145 0.67 0.73 1.95 3.85
pattern 8 −++−+− 9525 0.69 0.70 1.97 3.68
pattern 0 000000 9983 0.68 0.71 1.95 3.68
Collect important performance information to justify acceptable method
performance – aids in setting meaningful system suitability
criteria
19IFPAC 2014M. Argentine
Wavelength Robustness
Evaluate response for each
impurity to assess
robustness as
structural differences may cause
spectral response
differences
Ref: J. Chrom A, 762 (1997), 227-233
20IFPAC 2014M. Argentine
Wavelength Robustness
Ref: J. Chrom A, 762 (1997), 227-233
Indeed, different spectral profiles exist for the
various compounds (with need for ~280 nm
detection)…
…and a simple wavelength system suitability sample can be
developed with commercially-available materials
21IFPAC 2014M. Argentine
Method Transfers and Maintenance Method transfers should include:
• An assessment for capability to perform the analysis AND
• The process and product knowledge and overall control strategy for intended use of the method
Important for proper method use and for reference in any future improvements
Useful tools to assess performance/variability:
• Data from robustness studies to interpret system suitability results
• Data from method performance (control) samples in development aid in confirming appropriate performance during transfer and beyond.
Transfer of knowledge regarding “investigational” methods (e.g., broader-based gradient methods, spectroscopic studies) is also useful for studying potential impact of future process design space changes relative to the analytical design space
• Example: Evaluation of new sources of starting materials or excipients22IFPAC 2014
M. Argentine
Performance data allows trends to be identified (and understood)
0.50
0.70
0.90
1.10
1.30
1.50
0 20 40 60 80 100
data point
peak
taili
ng
0.50
0.70
0.90
1.10
1.30
1.50
100 120 140 160 180
data point
peak
taili
ng
Then…
…and more recently
(Tailing limit NMT 1.5)
Change in instrumentation affected performance
New instrumentation
23IFPAC 2014M. Argentine
Opportunities
With potential manufacturing changes…
If studies to define the knowledge space are well captured, one can leverage historical knowledge and investigational methods to study continuous improvement opportunities and either re-confirm suitability of existing controls or recognize potential concerns
• Example: use of gradient HPLC investigational method versus isocratic method to ensure appropriate evaluation for potential late-eluting impurities, if anticipated or possible
Use appropriate chromatographic and spectroscopic tools to assess post-approval process changes for method impact.
24IFPAC 2014M. Argentine
QbD/Analytical Lifecycle Summary
Advantages• Acknowledges control strategy is integrated with both process and
analytical understanding and controls• Effectively aligned with information/knowledge generation;
possibilities to leverage a wide variety of tools for effective process understanding
• Logical progression throughout development with evolution of method focus to meet varying “customer” demands
Disadvantages• Analytical methods used in development not necessarily same as for
routine control– Possibly more method development time– Possible utilization of more and varied analytical tools– More method “bridging” considerations
25IFPAC 2014M. Argentine
Additional Messages and Parting ThoughtsThe QbD/ analytical lifecycle approach supports a methodical, integrated, requirements-based approach to method development and implementation
From an industrial perspective, the knowledge, design, and control spaces are studied relative to the specific process/formulation that is being commercialized.
This approach delivers robust process and analytical understanding and controls which focus on the selected process
Analytical method development from a lifecycle perspective is performed in two parts:
• Defining and developing the overall product control strategy and method requirements• Method definition, using well-selected control parameters (system suitability) and
method performance samples can be used to ensure quality of future results. Method robustness and modeling tools help to define meaningful conditions and controls.
Analytical control methods are tailored for the selected combination of process and analytical controls for that process
Therefore, one needs to assess a compendial method for suitability with the intended manufacturing process
26IFPAC 2014M. Argentine
A few references• Olsen, B.; Gavin, P.; Wozniak, T Quality by Design Considerations for
Analytical Methods AAPS News Magazine (2007),10(12), 16-22.
• Argentine, M.; Olsen, B.; Owens, P. Strategies for the Investigation and Control of Process-Related Impurities in Drug Substances, Adv. Drug Delivery Reviews (2007), 59(1), 12-26.
• Gavin, P.; Olsen, B. A Quality by Design Approach to Impurity Method Development for Atomoxetine Hydrochloride (LY139603), Journal of Pharmaceutical and Biomedical Analysis (2008), 46, 431-441.
• Gavin, Peter F.; Olsen, Bernard A.; Wirth, David D.; Lorenz, Kurt T. A quality evaluation strategy for multi-sourced active pharmaceutical ingredient (API) starting materials. Journal of Pharmaceutical and Biomedical Analysis (2006), 41(4), 1251-1259.
• Wirth, D. D.; Olsen, B. A.; Hallenbeck, D. K.; Lake, M. E.; Gregg, S. M.; Perry, F. M. Screening methods for impurities in multi-sourced fluoxetine hydrochloride drug substances and formulations.Chromatographia (1997), 46(9/10), 511-523.
27IFPAC 2014M. Argentine
Thank You!
Acknowledgements:
Bernie Olsen Steven Baertschi
Tim Wozniak Eric Jensen
Peter Gavin John Stafford
Zhenqui Shi Gordon Lambertus
Robert Forbes
28IFPAC 2014M. Argentine
backups
2/10/2014File name/location
Company ConfidentialCopyright © 2000 Eli Lilly and Company
29
Use of ICP-MS to confirm absence of “concerning” elemental impurities for simplified control
30
Pd (µg/g) As (µg/g) Cd (µg/g) Hg (µg/g) Pb (µg/g)
Range of results (n=27 batches)across development, scale-up and
validation
<0.005to
5.20All <0.005 All <0.005 All <0.005
<0.005to
0.198
Oral PDEa, µg/day 100 1.5 25 5 15
Limit in API assuming <20 mg/day MDDb, µg/g 5560 83 1390 278 833
Limit in API assuming a 10 g/day dose, µg/g 10 0.15 2.5 0.5 1.5
a Permissible Daily Exposureb Maximum Daily Dosage
• Most elements are <1/100th of safety limits (even conservative limits)• Only Pd, which was intentionally used in the manufacturing process was >30% of
the most conservative safety limit (and only in an early development batch)Control Pd routinely to NMT 10 ppm using robust, widely available ICP-OES approachJustify non routine testing/control for other elemental impurities
IFPAC 2014M. Argentine
Another example – Use of robustness studies to define useful system suitability
31
Method developed to resolve many potential impurities (a); only a few
impurities typically seen (b)
Reference: P.F. Gavin and B.A.Olsen, J. Pharm. Biomed. Anal., 46 (2008), pp 431-441.
IFPAC 2014M. Argentine
Robustness study through design of experiments
32
• Fractional factorial design for 5 variables with 4 center points 16+4 = 20 runs
• Tracked resolution of key resolution pairs in addition to run time and tailing data with carefully selected crude sample (contained key impurities)
IFPAC 2014M. Argentine
Robustness Study Prediction Profile
33
Results support robustness to buffer conc. and pH; note
runtime and pressure changes with n-propanol and T changes
Rs (1-2) is correlated with Rs changes for other peak pairs and is sensitive to n-propanol
and T changes
Impurities 1 and 2 are good candidates for meaningful
system suitability assessment of method separation performance
• Impurity 2 is commercially-available • Impurity 1 can be procured or easily
formed with in-situ degradation
IFPAC 2014M. Argentine
In-situ generation of meaningful system suitability sample
34IFPAC 2014M. Argentine
Method change and comparability – an example
35
Situation: A method change was desired after manufacturing/process change to also detect a potential new impurity (Impurity 7)
(Gradient method could detect impurity 7 but isocratic impurity method 1 could not.)
Result: Changed both Assay and Impurities methods to new methods that were more appropriate and capable for the new manufacturing process.
IFPAC 2014M. Argentine
mV
50.00
51.00
52.00
53.00
54.00
55.00
56.00
57.00
58.00
59.00
60.00
Minutes0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00
1 2 34 5 6
7
A
B
Method selectivity of
gradient:
A= unspikedmatrix
B=matrix with 0.05%
spike for impurities
1,2,6,7,8 and impurity 5 spiked at
0.15%
36
8
Gradient Impurity Method
IFPAC 2014M. Argentine
85.000
82.700
80.400
78.100
75.800
73.500
71.200
68.900
66.600
64.300
62.0000 120 240 360 480 600 720 840 960 1080 1200
TIME (SECONDS)
SYS SUIT
LOT G
LOT E
LOT D
LOT C
LOT B
LOT A
BLANK
Impu
rity
1 (D
I)
Impu
rity
2 (D
I)
Analytical Profile – Method 1
Impu
rity
3 (D
I)Im
purit
y 4
(PI)
Impu
rity
5 (P
I)
Impu
rity
6 (D
I)
37
Impu
rity
7 (P
I)
IFPAC 2014M. Argentine
38
Analytical Profile – Method 2m
V
52.00
54.00
56.00
58.00
60.00
62.00
64.00
66.00
68.00
70.00
72.00
74.00
76.00
78.00
80.00
Minutes0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 20.00 22.00 24.00 26.00 28.00 30.00
blank
Lot 1
Lot 2
Lot 3
Lot 4
System suitability
Impu
rity
7
Impu
rity
5
Impu
rity
4
Impu
rity
6Im
purit
y 2
Impu
rity
1
Impu
rity
3
New method can see all of the impurities of the previous method as well as the new method
IFPAC 2014M. Argentine
39
Method Comparison - ImpuritiesR
esul
ts
0.22
0.24
0.26
0.28
0.3
0.32
0.34
0.36
A B
Method
Oneway Analysis of Results By Method
A = Method 1
B = Method 2
Evaluation of one “impurity rich” sample by both
methods
No statistical difference between methods
IFPAC 2014M. Argentine
40
Method Comparison - Assay
Lot 2
Lot 1
97
98
99
100
101
102
103
034J
D4
resu
lt
12 24 36 48 60 72 84 96 108 132 156Sample
B01391 B10370
97
98
99
100
101
102
103
034J
D4
resu
lt
12 24 36 48 60 72 84 96 108 132 156Sample
B01391 B10370
97
98
99
100
101
102
103
034J
D4
resu
lt
12 24 36 48 60 72 84 96 108 132 156Sample
B01391 B10370Method 1 Meth. 2
Lot 1
resu
lt
97.598.098.599.099.5
100.0100.5101.0101.5102.0
019J
D0
Res
ult
5 10 15 20 25 30 35 40 45 50 55 60Sample
B01391 B10370
97.598.098.599.099.5
100.0100.5101.0101.5102.0
019J
D0
Res
ult
5 10 15 20 25 30 35 40 45 50 55 60Sample
B01391 B10370Method 1 Method 2
Lot 2
resu
lt
IFPAC 2014M. Argentine
41
Method Comparison01
9JD
0 R
esul
t
98
98.5
99
99.5
100
100.5
101
B01391 B10370
Method
1 2
Lot 1
Res
ult
No practical or statistical difference
between the two methods
IFPAC 2014M. Argentine