Transcript Trevor_WCBP poster 1_9_2015

Spectral Method for Quantitation of Color in Protein Drug Solutions
Jian Yin, Tom Patapoff, Travis Horst, Kimia Rahimi, Inna Notkin, Jian Zhang, Joseph Marhoul, Sue Skieresz, Bruce Kabakoff, Sarah Du and
Trevor Swartz (Genentech Inc.)
Gordon Leggett and Paul Barnes (HunterLab)
Transmission is transformed to BLUE, GREEN and RED values
using illuminant and human observer tristimulis values
Introduction
Vitamin B2 Transmission
In biotechnology industry, a visual method is mostly utilized for color
characterization of liquid drug product solutions. In this method an analyst
visually determines the color of the sample by choosing the closest match to
standard color series. This method can be subjective due to the requirement
of an analyst to make a judgment of the best match of color of the sample to
the standard color series. Furthermore, variability in preparation of the color
series solutions exists. In addition, the current method does not capture data
on hue and chroma that would allow for improved product characterization
and the ability to detect subtle differences between samples. To overcome
these challenges, we describe a quantitative method for color determination
that greatly reduces the variability in measuring color and allows for a more
precise understanding of color differences. Following CIE color industry
standards, this method converts a protein solution’s visible absorption
spectra to L*a*b* color space. Color matching is achieved within the L*a*b*
color space. We describe here the algorithm used such that the quantitative
method correlates with the current visual method. In addition we provide the
L*a*b* values for the EP color standards required for the Visual color
assessment method. We have determined the L*a*b* values of the EP color
standards by gravimetrically preparing and measuring multiple lots of the
color standards. We demonstrate the correlation between the visual and
quantitative method using both low and high concentration antibody
solutions.
Solar Spectrum
Take the absorption spectrum and transform to L*a*b* values
35
30
X
BY1
Relative
intensity
-15
Wavelength (nm)
500
600
Tristimulis
1
Protein
solution
0
Tristimulis value
400
500
600
0
400
700
=
500
600
0
400
700
500
600
0.5
Yellow 4
Green
Blue
Red
Green
Yellow 4
0
400
500
600
0
400
700
500
600
0
400
700
500
600
700
a
Protein solution assigned to yellow series, less intense than Y3
100
Yellow
200
Quantitative method correlates to visual assessment method with
improved precision
Sample ID
R1
mAb 1
mAb 2
mAb 3
mAb 4
mAb 5
mAb 6
mAb 7
mAb 8
mAb 9
mAb 10
mAb 11
mAb 12
mAb 13
mAb 14
mAb 15
R7
800
b*
1000 Green
L
Red
Water
1400
1800
Blue
0
-110 200
400
600
300
100
% Transmission
Absorption (OD)
1.4
R1
400
500
600
Wavelength (nm)
700
Yellow
Brown
•
•
•
•
•
Wavelength (nm)
Transmission spectrum Vitamin B2
HunterLab
report color (≤)
4
B7
B8
B8
Y4
B6
Y4
B5
B6
B5
B5
B6
BY1
B5
B4
BY4
5
B6
B9
B6
Y4
B5
BY5
B5
B5
BY4
B5
Y5
Y1
BY4
BY4
BY4
1
B6
B9
B6
B5
B5
B5
R6
B5
B5
B5
B5
BY1
B5
B4
B4
Locking L*a*b* values for EP color standards
•
800
900
1000
1100
1200
1300
a*
Color series is all within the ‘yellow quadrant’
R7, B7, BY7,Y7, G7 are all tightly clustered correlates with difficulty
in differentiating when compared to actual color standards
Larger spacing are found between more intensely colored samples
Color space is uniform in perception
Correlation of instrument readout with visual assessment for
samples with little color
80
• Allowed Hue and Chroma will
match visual assessment
method
• Less colored than B6 but very
Red or very Green hue is
allowed
• Less colored than R7 reports to
the Brown color series
60
40
20
400
Ref II
Ref I
Ref II
Ref I
Ref III
Ref II
Ref I
Ref IV
Ref I
Ref II
Ref II
Ref III
Ref III
Ref II
Ref II
EP visual assessment report color (≤)
Analyst ID
1
2
3
B7
BY6
B7
B9
B9
B9
B7
BY6
B7
BY4
Y4
BY4
B6
Y5
B6
BY5
Y5
BY5
B6
B6
B6
B6
Y5
B6
BY5
BY5
B5
B5
B5
B5
BY5
BY5
BY5
Y1
BY2
BY1
BY5
Y5
B5
BY3
BY4
BY3
B5
BY4
B4
L*a*b* values for EP color standards have been locked and will be published:
900
0
0
2000
110
20
22
25
50
55
50
60
125
125
150
150
150
160
200
206
Turbidity
• Visual assessment and quantitative method correlate for both high and low
concentration protein solutions (low and high turbidity)
• There is a slight bias of quantitative assessment method towards Brown color
series vs. Brown Yellow for visual assessment
 This is partially attributable to light scattering that occurs from light entering the
sides of the sample tube.
 This is demonstrated by altering visually observed data when tube is wrapped
in foil (data not shown).
• Quantitative method has much higher precision as compared to visual assessment
(see associated poster; ”Accuracy and Precision Assessment of a
Spectrophotometric Method for Quantitative Measurement of Color in Protein Drug
Solutions”)
Red
700
Absorption spectrum Vitamin B2
0.2
1800
800
Absorption can be converted to transmission (aligns with our
perception)
0.4
1600
Conc.
mg/mL
700
R7
0.6
1400
1000
Wavelength (nm)
0.8
Green
Yellow
Brown
Yellow
600
700
1
1200
b*
R7
1.2
1000
500
Absorption (OD)
Absorption (OD)
400
600
a*
100
200
500
800
Quantitative method describes relative perception of color
standards
1
Y 3.8
1
1
1200
Absorption of red color series
Y 3.6
Green
Yellow 3
b
700
Y 3.4
Wavelength (nm)
2000
400
15
Y 3.2
1
-110
0
a
10
Yellow 3
1
Challenges: Current visual color assay has inherent variability
0.5
5
700
1600
Brown
Brown Yellow
Yellow
Green Yellow
Red
0
Wavelength (nm)
• Color (hue) readout: An analyst picks the best match of sample to a color series
• Intensity readout: An analyst picks the next darkest color standard as test result
Absorption of a sample from each color series
-5
X
Yellow
quadrant
Basis for quantitative color method: Information about color is
captured in the absorption spectrum of color standards
-10
Shortest distance correlates to best match of hue and chroma (intensity) in
visual assessment method
400
• Variability in making up standard color solutions
• Subjectivity in picking best match color
Protein
solution
5
50
0
400
Vitamin B2
Current color measurement:
Red
15
0
110
Brown Yellow BY7
b
100
600
B9
20
Riboflavin observed through daylight illuminant
BLUE, GREEN and RED values transformed to visually uniform
L*a*b* color space
Brown
Brown
Green
Yellow
10
Wavelength (nm)
Protein
solution
test
sample
Yellow Brown
Yellow
25
Protein
solution
test
sample
Color characterized by a visual determination of color using EP
color standards
B1
Implementation of a quantitative color determination method for a
protein solution
500
600
Wavelength (nm)
700
b*
B6
• Absorption spectrum is mathematically converted to Transmission
Note: OD of 1 is equal to 10% Transmission
• Color vision initiated with blue, green and red photoreceptors
R7
a*
water
Gravimetrically prepared and measured L*a*b* values of 5 batches of each
color standard
• Variability of 5 batches of color standards measured by multiple analysts is
less than visually discernable (delta E<0.5)
• Statistically robust method used to lock L*a*b* values
Implementation of a quantitative color test method for GMP
testing of products (COA and stability)
• HunterLab Software has been customized such that the readout correlates
to the current visual assessment method
• Developed daily system suitability protocol
• Developed validation protocol
• Goal is to be operational in GMP environment in 2015
• Quantitative method report is consistent with visual assessment method with
improved precision
• Sample L*a*b* values will be recorded to allow for tracking /trending of color
(Lot to Lot variability and on stability)