A quantitative spectral method has been developed to precisely measure the color of protein solutions. In this method, a HunterLab UltraScan Vis spectrophotometer is utilized for capturing the visible absorption spectrum of a protein solution, which can then be converted to color values (L*a*b*). These quantitative values (L*a*b*) allow for calculating the best match of a sample’s color to an EP reference color solution. In order to qualify this instrument and assay for IMP QC use, an assessment was conducted to evaluate the system suitability and assay accuracy and precision. Standards for system suitability testing were developed in preparation for moving this assay into a GMP testing environment. Different instruments, cuvettes, protein solutions and analysts were compared in this study and the instrument accuracy, repeatability and assay precision were determined. The instrument and assay are found suitable for use in measuring color of drug substances and drug products. The quantitative spectral method is comparable to the current EP visual method, with significantly improved reproducibility. These studies demonstrate that this assay is suitable for use in a GMP laboratory for lot release and stability testing.

Accuracy and Precision Assessment of a Spectrophotometric Method for Quantitative Measurement of Color in Protein Drug Solutions

Accuracy and Precision Assessment of a Spectrophotometric Method for Quantitative Measurement of Color in Protein Drug Solutions

Jian Yin, Bruce Kabakoff, Tom Patapoff, Bartolo Chen, Joseph Marhoul, Norman Shih, Jian Zhang, Trevor Swartz and Kimia RahimiGenentech Inc.

Jian Yin, Bruce Kabakoff, Tom Patapoff, Bartolo Chen, Joseph Marhoul, Norman Shih, Jian Zhang, Trevor Swartz and Kimia RahimiGenentech Inc.

ASSAY ACCURACY

ASSAY PRECISION : INTERMIDIATE PRECISION

Wavelength accuracy check:•Use a didymium solid color standard to verify the wavelength accuracy at two specific wavelength: 430 nm and 570 nm. Photometric response check:•Use a neutral density filter to verify the midrange photometric response over the whole visible range (360 nm-780 nm)

Target values  EP Visual assessmentHunterLab

Spectral method

BY1 BY1 BY1.0

B1 B1 B1.0

B5 B5 B5.0

R5 R5 R5.0

GY4 GY4 GY4.0

BY4 BY4 BY3.9

The spectral method correlates well with the EP visual assessment method.

• Both the spectral method and the visual assessment method matched the target color values when using EP color reference solutions as testing samples

• The acceptance criteria is set at Max ΔE*2000≤0.5• The Max ΔE*2000 between each measurement to the

mean of the replicates was used as a single matrix to evaluate the assay variability

• The smaller the Max ΔE*2000, the smaller the assay variability

• Human eyes can only note the color differences when the ΔE*2000≥1

• The assay precision was determined using three representative protein solutions with different concentrations and turbidities. (Mab 3, Mab 5 and Mab 14 in above table)

Sample ID Mab 3 Mab 5 Mab 14

Reading # L* a* b*

color (≤) L* a* b*

color (≤) L* a* b*

color (≤)

1 99.14 -0.14 1.87  B6 97.87 -0.21 5.16 B5 97.42 -0.59 8.30  B4

2 99.24 -0.14 1.86 B6 97.92 -0.20 5.15 B5 96.95 -0.55 8.34 B4

3 99.30 -0.14 1.89 B6 97.95 -0.22 5.17 B5 97.41 -0.57 8.32 B4

4 99.16 -0.15 1.88 B6 98.03 -0.22 5.17 B5 97.39 -0.57 8.33 B4

5 99.04 -0.13 1.86 B6 97.93 -0.23 5.14 B5 97.32 -0.57 8.32 B4

6 99.35 -0.14 1.88 B6 98.05 -0.23 5.16 B5 97.52 -0.58 8.31 B4

Max ΔE*2000 0.03 0.04 0.05

Max ΔE*2000 ≤0.16

Instrument to instrument variability-Max ΔE*2000 ≤0.17

Sample ID

Mab 3 Mab 5 Mab 14

Reading #

L* a* b*Color

(≤)L* a* b*

Color (≤)

L* a* b*Color

(≤)

1 99.41 -0.13 1.90 B6 98.07 -0.22 5.12 B5 97.53 -0.58 8.33 B4

2 99.33 -0.13 1.93 B6 98.05 -0.22 5.12 B5 97.54 -0.57 8.32 B4

3 99.34 -0.13 1.91 B6 98.07 -0.21 5.14 B5 97.31 -0.58 8.29 B4

4 99.34 -0.13 1.91 B6 97.92 -0.23 5.08 B5 97.57 -0.58 8.31 B4

5 99.34 -0.13 1.91 B6 98.05 -0.22 5.14 B5 97.61 -0.58 8.31 B4

6 99.37 -0.14 1.89 B6 98.08 -0.23 5.16 B5 97.54 -0.55 8.38 B4

Max ΔE*2000 0.15 0.09 0.16

Instrument ID #1 #2 Max ΔE*2000

Sample ID L* a* b* L* a* b*  

Mab 3 99.36 -0.13 1.91 99.40 -0.15 1.90 0.08Mab 5 98.04 -0.22 5.13 98.10 -0.24 5.16 0.11

Mab 14 97.52 -0.57 8.32 97.54 -0.61 8.36 0.17

Analyst ID #1 #2 Max ΔE*2000

Sample ID L* a* b* L* a* b*  

Mab 3 99.21 -0.14 1.87 99.36 -0.13 1.91 0.23Mab 5 97.96 -0.22 5.16 98.04 -0.22 5.13 0.14

Mab 14 97.34 -0.57 8.32 97.52 -0.57 8.32 0.30

Cuvette 3.5 mL standard cuvette 1.4 mL semi-micro cuvette

Reading # L* a* b* color (≤) L* a* b* color (≤)

1 98.75 -0.34 4.91 B5 98.49 -0.29 5.01 B5

2 98.76 -0.34 4.90 B5 98.18 -0.31 4.91 B5

3 98.48 -0.33 4.91 B5 98.45 -0.28 4.98 B5

Average 98.66 -0.34 4.91   98.37 -0.29 4.97  

SD 0.16 0.01 0.01   0.17 0.02 0.05  

• The HunterLab UltraScan Vis spectrophotometer is suitable for quantitatively measuring the color of a protein solution.

• The spectral measurement of color solutions independent of protein concentration and/or turbidity.

• A didymium standard and a 0.2 OD neutral density filter are suitable for the system suitability check.

• The quantitative spectral method is comparable to the current EP visual assessment method, but can produce more precise color values.

• The precision of the spectral assay is acceptable compared to that of the visual assessment..

• The qualitative spectral method is suitable for using in a GMP environment.

ACKNOWLEDGMENT

Margaret Tang, Travis Horst, Inna Notkin and Samir Sane

ASSAY PRECISION : ACCEPTANCE CRITERIAS

CONCLUSIONS

ASSAY PRECISION : ASSAY REPEATABILITY

MINIMIZING SAMPLE VOLUME

ASSAY PRECISION : INSTRUMENT REPEATABILITY

SYSTEM SUITABILITY

INTRODUCTION

Max ΔE*2000 ≤0.05.

Sample ID Conc. Turbidity COC report color (≤) HunterLab report color (≤) mg/mL Analyst ID Analyst ID 1 2 3 4 5 1 2 3 4 5

Mab1 20 Ref II B7 BY6 B7 B7 B6 B6 B6 B6 B6 B6Mab2 22 Ref I B9 B9 B9 B8 B9 B9 B9 N/A B9 B9Mab 3 25 Ref II B7 BY6 B7 B8 B6 B6 B6 B6 B6 B6Mab4 50 Ref I BY4 Y4 BY4 Y4 Y4 B5 B5 B5 B5 B5Mab 5 50 Ref III B6 Y5 B6 B6 B5 B5 B5 B5 B5 B5Mab 6 50 Ref II BY5 Y5 BY5 Y4 BY5 B5 B5 B5 B5 B5Mab 7 60 Ref I B6 B6 B6 B5 B5 R6 R6 R6 R6 R6Mab 8 125 Ref IV B6 Y5 B6 B6 B5 B5 B5 B5 B5 B5Mab 9 125 Ref I BY5 BY5 B5 B5 BY4 B5 B5 B5 B5 B5

Mab 10 150 Ref II B5 B5 B5 B5 B5 B5 B5 B5 B5 B5Mab 11 150 Ref II BY5 BY5 BY5 B6 Y5 B5 B5 B5 B5 B5Mab 12 150 Ref III Y1 BY2 BY1 BY1 Y1 BY1 BY1 BY1 BY1 BY1Mab 13 160 Ref III BY5 Y5 B5 B5 BY4 B5 B5 B5 B5 B5Mab 14 200 Ref II BY3 BY4 BY3 B4 BY4 B4 B4 B4 B4 B4Mab 15 206 Ref II B5 BY4 B4 BY4 BY4 B4 B4 B4 B4 B4

• Variable result between different analysts by the visual assessment method

• Consistent result between different analysts by the spectral method

• Small bias between the two methods when using protein solutions as testing samples which is partially attributable to the light scattering effect on protein solutions

Analyst to analyst variability-Max ΔE*2000 is ≤0.30.

• The 3.5 mL standard clear wall cuvette (10 x 10 mm) and the 1.4 mL (4 X 10 mm) black wall semi-micro cuvette with 1 cm path length can produce comparable and reproducible result thus are recommended to use.

• The 3.5 mL disposable acrylic cuvette (10 X 10 mm) with 1 cm path length can be used for ADC and potent protein solutions (data not shown).

• The 0.7 mL black wall cuvette (4 X 5 mm) with 1 cm path length can not produce reproducible result, thus can not be used (data not shown).

• One analyst measured each sample 6 times on the same instrument without sample refills

• One analyst measured each sample 6 times on the same instrument with sample refills

• One analyst measured each sample 6 times on each of the two instruments

• Two analysts measured each sample 6 times on the same instrument