Process Analytical Technology for assessment and control of API content uniformity Available Technologies Steve Hammond Pfizer Global Manufacturing
Process Analytical Technology for
assessment and control of API
content uniformity
Available Technologies
Steve Hammond
Pfizer Global Manufacturing
Presentation Outline• Standard At-line application
• Batch blender system
• Continuous tablet core measurements
• Continuous Processing PAT systems
• Blend in a transfer chute
• Future RTR measurement strategy
• Potential for feed frame measurements
• Conclusions
1030 1040 1050 1060 1070 1080 1090 1100 1110 1120 1130 11401/0.0001
-3.1854
-2.6869
-2.1884
-1.6899
-1.1915
-0.6930
-0.1945
0.3040
0.8025
1.3010
1.7995
5mg: Loadings
Wavelength/nm
Inte
nsit
y X
10^4
10mg Calibration Regression Line
y = 0.9915x + 0.8313
R2 = 0.9915
50.00
60.00
70.00
80.00
90.00
100.00
110.00
120.00
130.00
140.00
50.00 70.00 90.00 110.00 130.00 150.00
%Intent by HPLC
%In
ten
t b
y N
IR
3
NIR for At-line CU Determination• Conventional lab-based NIR system
– Uses discriminating spectral signature of API
– Validated examples over range 0.25 – 75%
– Can be compressed blend or tabletsBlack Active
Red placebo
Aromatic C-H
NIR_Intact/MS/24Nov08 4
Capsule NIR Transmission SystemSource
Detector
NIR_Intact/MS/24Nov08 5
PLS Regression - API
912 nm 1156 nm
3 Factors selected
(94% of X variance
explained)
Very little
unrelated
spectral
variance within
the data set
1018 nm
1350 nm
API & Excipients
Dispensing Blend BlendSievingDispensing Blend BlendSieving Granulation
Mag
Stearate
Mag
Stearate
PAT1
PAT1
PAT2
PAT2
BlendTablettingCoating PAT3
PAT3
PAT4
PAT4
PAT2
PAT2
PAT2
PAT2
PAT:
1 = Identification
2 = Blend Uniformity
3 = Weight, Hardness, Disintegration, Potency, Content Uniformity
4 = Water Determination
Product “A” Real Time Release and PAT
NIR Mounted on a rotating blender
NIR in enclosure
Rotation point
-1
0
1
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
tPS
[1]
$Time (normalized)
Full_Set_SNV.M1:Step_1, PS-Full_Set_SNV_NEW_Jan
Predicted Scores [Comp. 1]
+3 Std.Dev
t[1] (Avg)
-3 Std.Dev
tPS[1] (Batch 810005500)
tPS[1] (Batch 810011300)
SIMCA-P+ 11.5 - 21/01/2008 12:44:22
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180
t[1]
$Time (normalized)
Full_Set_SNV.M1:Step_1
Scores [Comp. 1] (Aligned)
+3 Std.Dev
t[1] (Avg)
-3 Std.Dev
SIMCA-P+ 11.5 - 18/12/2007 10:54:09
-3
-2
-1
0
1
-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
t[1]
$Time (normalized)
Full_Set_SNV.M2:Step_2
Scores [Comp. 1] (Aligned)
+3 Std.Dev
t[1] (Avg)
-3 Std.Dev
SIMCA-P+ 11.5 - 18/12/2007 11:43:42
-0.2
-0.1
-0.0
0.1
0 10 20 30 40 50 60 70 80 90 100 110 120
t[1]
$Time (normalized)
Full_Set_SNV.M3:Step_3
Scores [Comp. 1] (Aligned)
+3 Std.Dev
t[1] (Avg)
-3 Std.Dev
SIMCA-P+ 11.5 - 18/12/2007 11:06:43
-5
-4
-3
-2
-1
0
1
0 10 20 30
t[1]
$Time (normalized)
Full_Set_SNV.M4:Step_4
Scores [Comp. 1] (Aligned)
+3 Std.Dev
t[1] (Avg)
-3 Std.Dev
SIMCA-P+ 11.5 - 18/12/2007 11:11:14
8
Control Charts
NIR interfaced with Press
10X Sampling Unit
NIR Process Analyzer
Report
MES-Systems
Process Analyzers• Measure condition of the process material in real time
• Collect more information about the batch
6.305.955.605.254.904.554.203.85
4000
3000
2000
1000
0
Potency, mg/g
Nu
mb
er
of
Un
its
85%75% 115% 125%
Mean 4.958
StDev 0.1058
N 17547
Normal
Histogram of Potency (mg/g) - Nominal 5 mg/g
Results from 18,000 tablets for a 5mg dose in a 200mg tablet
5.165.105.044.984.924.86
Median
Mean
5.0255.0205.0155.0105.0055.000
1st Q uartile 4.9670
Median 5.0089
3rd Q uartile 5.0554
Maximum 5.1910
5.0047 5.0234
4.9971 5.0198
0.0597 0.0730
A -Squared 0.45
P-V alue 0.268
Mean 5.0141
StDev 0.0657
V ariance 0.0043
Skewness 0.247446
Kurtosis -0.004509
N 192
Minimum 4.8372
A nderson-Darling Normality Test
95% C onfidence Interv al for Mean
95% C onfidence Interv al for Median
95% C onfidence Interv al for StDev
95% Confidence Intervals
Summary
Position of mean and median
106104102100989694
160
140
120
100
80
60
40
20
0
% Nominal
Fre
qu
en
cy
Mean 99.59
StDev 1.657
N 1150
Histogram % Nominal (52 Batches)
Distribution for 1140 capsules from 52 lots
Next Generation Manufacturing
Initiative in Pfizer
• An extension of Quality by Design to achieve
–Continuous manufacturing of a drug product
–Control Strategy - use Process Analytical Technology
(PAT) and Design Space to demonstrate control
–Process validation by Continuous Quality Verification
(CQV)
– Introduction of Real Time Release (RTR)
• Immediate release solid oral dosage form (capsule)
Schematic of Continuous Process
Encapsulation
PAT Measurement Point: Potency
PAT Measurement : Particle Size Distribution
Raw Material Feeders
Mean residence time 200 secs
Turbine speed 50rpm
Retained mass 2.5Kg
Throughput 50Kg/h
Primary mixing
Lubrication mixing
PAT Measurement Point: Potency
Process parameters determining
potency and CU• In a typical DP process:
Primary Blend Potency (CU) = weighing + mixing
• In a batch process weighing and mixing are a one time event.
• In a continuous process, weighing and mixing operation is repeated many
ten’s of thousands of times, the operation is time variant in nature.
• For a continuous process, the material flow can be considered a stream of
“mico batches”, each having a discrete potency and CU.
• Sampling of a continuous process should take into account the rate at which
“mico batches” are flowing from the system.
• Sampling should be time variant, and match material throughput
Continuous Dry Granulation
PAT Systems – Probe Interface
Proprietary PAT Probe Interfaces have been developed to minimize the impact of the dynamic powder flow and blend cohesive properties on the measurement capability
Final Blend Probes Interface
for NIR & FBRMFBRM Probe Interface after
Roller Compactor
Blend Uniformity Measurement by NIR
• Two measurement points:
– Initial blend
–Final blend
oMeasurement is in real time
• Measurement is under dynamic flow conditions
o A probe based NIR spectrometer is used
o A sampling interface is constructed
o Critical is an understanding of measurement
characteristics
o Contributing mass to a spectrum, scan speed.
Sample Size Calculation
The sample size that ePAT611 sees depends on
• Probe spot size
• Penetration Depth of light
• Powder Density
Probe
Powder Bed
Sample that probe sees
The sample size calculation assumed the following:• Acquisition time is 20ms, each spectrum is the average of 50 scans
• Probe spot diameter is 2mm,
• Depth of penetration is 0.5mm,
• Powder density is 0.7mg/mm3
Under static condition, the amount each scan sees can be estimated by:
M = p (d/2)2DH
where d is the probe spot diameter, D is the powder density, and H is the penetration depth
M = 1.1mg/scan
Scanning for 1 second and adding the spectra -The sample mass for each spectrum is approximately :
55mg
Calibration Model Building
• Calibration mixtures were prepared below and above
the nominal concentration
• Pre-blended material was fed through the continuous
process
• Spectra were taken approximately every second
• Spectra collected were divided into Calibration set and
Test set
• Multivariate Statistical techniques were used to develop
calibrations.
• Two confirmation runs were used to evaluate the model
performance
NIR Calibration Model
Predicted Potency
Ac
tua
l P
ote
ncy
Sample Number
Actual Potency
NIR Predicted
Potency
Predicted vs. Actual Potencies for Calibration dataset
Po
ten
cy +/- 12% Nominal
Nominal
NIR Calibration Model
Predicted Potency
Ac
tua
l P
ote
ncy
Sample Number
Actual Potency
Predicted Potency
Predicted vs. Actual Potencies for Internal validation dataset
Po
ten
cy
+/- 12% Nominal
Model PerformanceNIR predicted Potencies for two confirmation runs
targeting at nominal potency
Nominal Potency
Nominal Potency
- 10%
Nominal Potency
+ 10%The Average
Capsule Potency by
HPLC = 99.3%LC
Standard platform for RTR
• RTR has significant benefits to Pfizer
• To supply chain management
• Reduced testing in a laboratory
• Increased process understanding
• Whilst also increasing the level of Quality
Assurance of the product
• Require an easy to implement, low cost,
standard platform for the application of PAT
Schematic of Feed Frame PAT
Installation
Right Paddle wheel
direction of rotation
Empty
Tablet DiesDirection of Tablet Die
Movement
Left Paddle Wheel
direction of rotation
Powder Inlet
Chute
Full
Tablet
Dies
Example Probe InstallationsManesty Unipress Diamond (USA) and Fette 2090 (Germany)
SentroProce-PAT 611
The design and engineering of the probe interface is still on-going,
but expected to be complete by the end of 2011
26
Technology AdvancementsQuantitative Method Development – Norvasc
Example
85
90
95
100
105
110
115
0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500
Num
Observed vs. Prediticted Line Plot, e-PAT 611 YPred[2](%_LC)
YVar(%_LC)
SIMCA-P+ 12 - 2010-05-02 19:02:36 (UTC-5)
Norvasc blends premixed at 85%, 93%, 100%, 107%, and 115% LC
Norvasc is 3.5% API and a good
test case for measurement capability+ /- 15% of nominal
nominal
Potential Benefits
• Increased process understanding of blending and
compression processes.
• Understand and monitor feed-frame function
• Ability to detect segregation during powder transfer from
IBC to the tablet press
• Applicable to both Batch and Continuous Processes
• Integration of PAT signal and tablet press weight control
signal into compression machine logic.
– Advanced Process Control
• Opportunity to implement as part of RTRt paradigm
RTR Approach for Continuous
Processes
• ICH Q8 (R1) - “Unit dose uniformity performed
in-process (e.g., using weight variation coupled
with near infrared (NIR) assay) can enable real
time release testing and provide an increased
level of quality assurance compared to the
traditional end-product testing using
compendial content uniformity standards”
Question?
• With the development of fast continuous
measurement systems in the feed frame, is a
CU measurement still a relevant statistic?
• If yes how do you deal with 10,000’s of
results?
• Is it a “discrete” measurement approach
• Or a statistic designed for a continuum of
data?
Summary
• Conventional At-line NIR transition measurements
now common – tablets and capsules
• On-line tablet core analysis is providing good
capability data
• PAT is important for Continuous Manufacturing
• NIR methods have been developed for real time
monitoring of Blend Uniformities in a Continuous
Dry Granulation process.
• NIR capability is close for feed-frame measurement
of CU, and may become the standard approach
• The predicted NIR results will be used for Process
Control; Continuous Quality Verification and Real
Time Release of the final product.
Acknowledgement of the whole army
Acknowledgements
• Many, many people have contributed to this
presentation:
• Pfizer World Wide R&D: Groton Pharm Sciences
• Pfizer Global Supply: Process Analytical Support
Group
• Pfizer Global Supply: Technology, Science and
Operation Group
• Pfizer Global Supply: Caguas Site, Illertissen, Freiburg
• Pfizer Global Supply: Product and Process
Development, Freiburg