catalent overcoming assay drift and multiple matrix interference · pdf file ·...

1
® Catalent Reliable Solu ons. Inspired Results. *Presenting Author: [email protected] Research Triangle Park, North Carolina, USA Overcoming Assay Drift and Multiple Matrix Interference Challenges in a PK ELISA Simone S. Cummings*, Ph.D., Gwendolyn Wise-Blackman, Ph.D, Catalent Pharma Solutions, Biopharmaceutical Services, RTP, NC, USA Abstract Purpose: Absence of assay drift in ELISAs needs to be demonstrated by executing low mid and high QCs at the beginning and end of a 96-well plate, especially in the presence of complex biological matrices. We present a case-study demonstrating a Design of Experiment (DOE) approach to develop a sensitive ELISA to detect a therapeutic pegylated protein (TPZ) in monkey, rat and human serum that surmounts the challenge of assay drift and matrix interference. Validation results will be shown. Methods: Sheep anti-protein Z was coated onto the ELISA plates and pooled serum spiked with TPZ was used as standards and controls. The captured TPZ was exposed to biotinylated anti-protein Z antibody. The detection of the captured TPZ was performed with streptavidin labeled with horseradish peroxidase. To overcome the challenge of assay drift, we varied the incubation times of samples, used non-binding plates and a 12-channel pipette in transferring standards and QCs to binding plates and evaluated plates with different binding efficiencies. To overcome the additional challenge of matrix interference in human serum we used chaotropic salts. Results: Using these approaches we observed no matrix interference or assay drift in monkey, rat or human serum. The assay was accurate and precise in all matrices over a linear range of 39.05 – 10,000 ng/mL in rat, 78.15 – 7500 ng/ml in monkey and 50.00-10,000 ng/ml in human serum. The inter assay precision (%CV of mean back calculated results from two analysts) for two sets of low mid and high QCs ranged from 10.7 to 20.0 for monkey, 14.3 to 16.6 for rat serum and 9.3 to 24.7 for human serum. The intra-assay precision (%CV of the mean back calculated results of two sets of low mid and high QCs from one analyst) ranged from 0.3 to 13.3 for monkey, 0.2 to 7.9 for rat and 0.0 to 9.1 for human serum. Conclusion: We were able of surmount the challenge of assay drift and matrix interference and validate an ELISA for the detection of TPZ in monkey, rat and human serum that was highly sensitive, linear, accurate, precise and rugged. Conclusion • Increasing the coating time from 1 to 2 hours and using a higher binding 96-well plate reduced the assay drift observed in Cynomolgus monkey serum. • 10% DMSO reduced matrix effect in human serum. • MgCL 2 and MgCL 2 with DMSO drastically increased the C-value of the standard curve. • Increasing the incubation time for the primary and secondary anitbody did not improve assay drift. • Reducing the substrate incubation temperature to 4C and changing the orientation of our standards did not improve assay drift. Methodology Sheep anti- protein Z was coated onto the ELISA plates and pooled serum spiked with TPZ was used as standards and controls. The captured TPZ was exposed to biotinylated anti-protein Z antibody. The detection of the captured antibody was performed with streptavidin labeled with horseradish peroxidase HRP Substrate Blue product HRP Sheep anti-protein Z TPZ Biotinylated anti-protein Z Figure 1: Assay design Wash and incubate with samples for 1 hour Was and add primary antibody for 1 hour Wash and add secondary antibody for 1 hour Wash and add substrate for 10 minutes, stop and read at 450 Coat and block 96-well plate for 1 hour Wash and add substrate for 10 minutes, stop and read at 450 Wash and add secondary antibody for 1 hour Was and add primary antibody for 1 hour Wash and incubate with samples for 1 hour Method Overview Figure 2: Basic outline of assay method Assay Drift Quality control samples on the right side of the plate had > 20%RE. The back calculated concentrations were consistently lower than the nominal concentration 1 2 3 4 5 6 7 8 9 10 11 12 A STD 1 STD 2 STD 3 STD 4 STD 5 STD 6 STD 7 STD 8 STD 9 STD 10 STD 11 STD 12 B STD 1 STD 2 STD 3 STD 4 STD 5 STD 6 STD 7 STD 8 STD 9 STD 10 STD 11 STD 12 C HQC HQC D MQC MQC E LQC LQC F HQC HQC G MQC MQC H LQC LQC Figure 3: ELISA plate map showing the showing the layout of standards and quality control samples Varying Incubation Times -1 0 1 2 3 4 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Method 2 Hrs Coating 2 Hrs Primary Ab 2 Hrs Secondary Ab Log [Anti-proteinZ] OD Figure 4: Four parameter curve representing standards that were tested under various conditions Figure 5: Four parameter curves of standards treated with MgCL 2 and DMSO. Human Serum, DMSO and MgCl 2 Figure 6: Four parameter curves of standards treated with MgCL2. Human Serum and MgC 2 Assay Drift Across 96-Well Plate Quality control sample Concentration (ng/mL) Precision (%CV) Back-calculated concentration Relative error (%RE) LQC 1 10.0 5.6 8.8 11.5 MQC 1 50.0 6.4 46.0 8.0 HQC 1 135.0 9.0 144.2 6.8 LQC 2 10.0 1.1 5.7 43.0 MQC 2 50.0 4.2 33.1 33.7 HQC 2 135.0 2.7 96.2 28.7 Table 1: The concentration, precision and relative error of quality control samples. One aliquot of each QC sample was placed on the left side of a 96 well plate ( 1 wells D 1 and 2, F 1 and 2 and E 1 and 2). The second aliquot was placed on the right side of the same plate ( 2 wells F 11 and 12, H 11 and 12 and G 11 and 12). QCs that Pass QCs that Fail Standards Demonstrating Assay Drift Table 2: A comparison of standards in various orientation on a 96-well plate. Standard 1 was placed in columns 1 and 2 and compared to standard 2 which was plated in columns 11 and 12. The plate was developed at 4 C. Standard concentration (ng/mL) Standard 1 mean value Standard 2 mean value Mean % difference 1600 2.495 2.592 -4 800 2.520 2.582 -2 200 2.202 2.298 -4 100 1.918 1.991 -4 50 1.495 1.553 -4 25 1.065 1.085 -2 12.5 0.636 0.722 -14 3.125 0.242 0.283 -17 Varying Incubation Times Best-fit values One hour coating, 1 hour 1 , 1 hour 2 **Two hours coating, 1 hour 1 , 1 hour 2 One hour coating, *2 hour 1 , 1 hour 2 One hour coating, 1 hour 1 , *2 hours 2 Bottom 0.135 0.137 0.175 0.192 Top 2.779 3.334 2.871 2.870 Hillslope 1.037 1.025 1.081 1.206 EC50 50.53 48.57 43.93 57.5 R 2 0.998 1.000 0.998 1.000 Sig/Bkg 20.65 24.41 16.42 14.94 Table 3: **Incubation times that resulted in the best signal to noise ratio. *Original incubation time increased. Quality control sample Concentration (ng/mL) Total number of samples tested in validation Precision Number of samples < 20 %CV Relative error - number of samples < 25 %RE LQC 1 10.0 8 8 8 MQC 1 50.0 8 8 8 HQC 1 125.0 8 8 8 LQC 2 10.0 8 8 7 MQC 2 50.0 8 8 8 HQC 2 125.0 8 8 6 Accuracy and Precision Validation Data Table : Validation data showing the accuracy and precision of quality controls in Cynomolgus monkey serum. 4 Matrix Interference Observed in Individual Serum Serum lot number Concentration (ng/mL) Precision (%CV) Back-calculated concentration Relative error (%RE) 298057 125.0 9.5 55.89 55.3 10.0 8.4 6.79 32.1 298060 125.0 1.6 58.81 53.0 10.0 6.4 7.25 27.5 298055 125.0 3.5 79.61 36.3 10.0 1.6 5.1 48.8 Table 5: 4% Human serum spiked at the high and low quality control concentrations. Serum lot number Concentration (ng/mL) Precision (%CV) Back-calculated concentration Relative error (%RE) 298057 125.0 3.5 138.5 10.8 10.0 3.4 9.58 4.2 298060 125.0 1.6 143.26 14.6 10.0 6.4 9.51 4.9 298055 125.0 3.5 148.0 18.4 10.0 1.6 10.14 1.5 Serum Treated with 10% DMSO Table 6: 4% Individual lots of human serum spiked at the high and low quality control concentrations. Matrix Effect Validation Data Matrix Lot number LQC Met %RE ≤ 25% %RE HQC Met %RE≤25 %RE Human serum 298035 Yes 19.0 Yes 18.7 298037 Yes 19.7 Yes 5.1 298038 Yes 19.0 Yes 15.8 298057 No 28.6 Yes 20.2 298058 Yes 19.1 Yes 4.1 298064 No 31.4 Yes 24.9 Table 7: Minimal matrix effect observed in 2% individual human serum during validation. Serum was not treated with MgCL 2 of DMSO. Correcting Assay Drift • Add standards, QCs and samples to non-binding plate before transferring to binding plate with a 12-channel pipette. •*Used a 96-well ELISA plate with a higher binding efficiency. •Evaluated various orientation of standards and QCs. •Evaluated various substrate incubation temperatures * Changing the 96-well plate to a plate with higher binding efficiency resulted in a drastic improvement of the assay drift. Matrix Interference • Six randomly selected individual human serum were tested for matrix interference. Serum were spiked and tested at the high and low quality control concentrations. Most individual serum had a back-calculated value that was lower than the nominal concentration and > 25% RE. Correcting Matrix Interference • Treat 4% human serum with 10% DMSO, 0.5M MgCL 2 + 10% DMSO and 1M MgCL 2 + 10% DMSO. Treat 4% human serum with 0.5M MgCL 2 , and 1M MgCL 2 . 4% individual serum lots were treated with 5% and 10% DMSO. Matrix Effect End Result • Matrix effect was drastically improved with 10% DMSO. The presence of MgCL 2 increased the c-value of the standard curve. More assays are required to complete the development of the method using DMSO and MgCL 2 . We were able to surmount the challenge of matrix interference by lowering our LLOQ and using a 1:2 dilution of human serum. l

Upload: doandang

Post on 28-Mar-2018

217 views

Category:

Documents


1 download

TRANSCRIPT

®CatalentReliable Soluti ons. Inspired Results.™

*Presenting Author: [email protected] Triangle Park, North Carolina, USA

Overcoming Assay Drift and Multiple Matrix Interference Challenges in a PK ELISASimone S. Cummings*, Ph.D., Gwendolyn Wise-Blackman, Ph.D,

Catalent Pharma Solutions, Biopharmaceutical Services, RTP, NC, USA

AbstractPurpose: Absence of assay drift in ELISAs needs to be demonstrated by executing low mid and high QCs at the beginning and end of a 96-well plate, especially in the presence of complex biological matrices. We present a case-study demonstrating a Design of Experiment (DOE) approach to develop a sensitive ELISA to detect a therapeutic pegylated protein (TPZ) in monkey, rat and human serum that surmounts the challenge of assay drift and matrix interference. Validation results will be shown.Methods: Sheep anti-protein Z was coated onto the ELISA plates and pooled serum spiked with TPZ was used as standards and controls. The captured TPZ was exposed to biotinylated anti-protein Z antibody. The detection of the captured TPZ was performed with streptavidin labeled with horseradish peroxidase. To overcome the challenge of assay drift, we varied the incubation times of samples, used non-binding plates and a 12-channel pipette in transferring standards and QCs to binding plates and evaluated plates with di� erent binding e� ciencies. To overcome the additional challenge of matrix interference in human serum we used chaotropic salts.Results: Using these approaches we observed no matrix interference or assay drift in monkey, rat or human serum. The assay was accurate and precise in all matrices over a linear range of 39.05 – 10,000 ng/mL in rat, 78.15 – 7500 ng/ml in monkey and 50.00-10,000 ng/ml in human serum. The inter assay precision (%CV of mean back calculated results from two analysts) for two sets of low mid and high QCs ranged from 10.7 to 20.0 for monkey, 14.3 to 16.6 for rat serum and 9.3 to 24.7 for human serum. The intra-assay precision (%CV of the mean back calculated results of two sets of low mid and high QCs from one analyst) ranged from 0.3 to 13.3 for monkey, 0.2 to 7.9 for rat and 0.0 to 9.1 for human serum. Conclusion: We were able of surmount the challenge of assay drift and matrix interference and validate an ELISA for the detection of TPZ in monkey, rat and human serum that was highly sensitive, linear, accurate, precise and rugged.

Conclusion• Increasing the coating time from 1 to 2 hours and using a higher binding 96-well plate reduced the assay drift observed in Cynomolgus monkey serum.

• 10% DMSO reduced matrix eff ect in human serum.

• MgCL2 and MgCL2 with DMSO drastically increased the C-value of the standard curve.

• Increasing the incubation time for the primary and secondary anitbody did not improve assay drift.

• Reducing the substrate incubation temperature to 4C and changing the orientation of our standards did not improve assay drift.

Methodology

Sheep anti- protein Z was coated onto the ELISA plates and pooled serum spiked with TPZ was used as standards and controls. The captured TPZ was exposed to biotinylated anti-protein Z antibody. The detection of the captured antibody was performed with streptavidin labeled with horseradish peroxidase

HRP

Substrate Blue product

HRP

Sheep anti-protein Z

TPZ

Biotinylated anti-protein Z

Figure 1: Assay design

Wash and incubate with samples for 1 hour

Was and add primary antibody for 1 hour

Wash and add secondary antibody for 1 hour

Wash and add substrate for 10 minutes, stop and read at 450

Coat and block 96-well plate for 1 hour

Wash and add substrate for 10 minutes, stop and read at 450

Wash and add secondary antibody for 1 hour

Was and add primary antibody for 1 hour

Wash and incubate with samples for 1 hour

Method Overview

Figure 2: Basic outline of assay method

Assay Drift

• Quality control samples on the right side of the plate had > 20%RE.

• The back calculated concentrations were consistently lower than the nominal concentration

1 2 3 4 5 6 7 8 9 10 11 12

A STD 1 STD 2 STD 3 STD 4 STD 5 STD 6 STD 7 STD 8 STD 9STD 10

STD 11

STD 12

B STD 1 STD 2 STD 3 STD 4 STD 5 STD 6 STD 7 STD 8 STD 9STD 10

STD 11

STD 12

C HQC HQC

D MQC MQC

E LQC LQC

F HQC HQC

G MQC MQC

H LQC LQC

Figure 3: ELISA plate map showing the showing the layout of standards and quality control samples

Varying Incubation Times

-1 0 1 2 3 40.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5Method2 Hrs Coating2 Hrs Primary Ab2 Hrs Secondary Ab

Log [Anti-proteinZ]

OD

Figure 4: Four parameter curve representing standards that were tested under various conditions

Figure 5: Four parameter curves of standards treated with MgCL2 and DMSO.

Human Serum, DMSO and MgCl 2

Figure 6: Four parameter curves of standards treated with MgCL2.

Human Serum and MgC 2

Assay Drift Across 96-Well PlateQuality control

sampleConcentration

(ng/mL)Precision

(%CV)Back-calculated concentration

Relative error (%RE)

LQC1 10.0 5.6 8.8 11.5

MQC1 50.0 6.4 46.0 8.0

HQC1 135.0 9.0 144.2 6.8

LQC2 10.0 1.1 5.7 43.0

MQC2 50.0 4.2 33.1 33.7

HQC2 135.0 2.7 96.2 28.7

Table 1: The concentration, precision and relative error of quality control samples. One aliquot of each QC sample was placed on the left side of a 96 well plate (1wells D 1 and 2, F 1 and 2 and E 1 and 2). The second aliquot was placed on the right side of the same plate (2wells F 11 and 12, H 11 and 12 and G 11 and 12).

QC

s that P

assQ

Cs that

Fail

Standards Demonstrating Assay Drift

Table 2: A comparison of standards in various orientation on a 96-well plate. Standard 1 was placed in columns 1 and 2 and compared to standard 2 which was plated in columns 11 and 12. The plate was developed at 4 C.

Standard concentration

(ng/mL)

Standard 1 mean value

Standard 2 mean value

Mean % difference

1600 2.495 2.592 -4

800 2.520 2.582 -2

200 2.202 2.298 -4

100 1.918 1.991 -4

50 1.495 1.553 -4

25 1.065 1.085 -2

12.5 0.636 0.722 -14

3.125 0.242 0.283 -17

Varying Incubation TimesBest-fit values

One hour coating, 1 hour 1 , 1

hour 2

**Two hours coating, 1 hour 1 , 1

hour 2

One hour coating, *2 hour 1 , 1

hour 2

One hour coating, 1

hour 1 , *2 hours 2

Bottom 0.135 0.137 0.175 0.192

Top 2.779 3.334 2.871 2.870

Hillslope 1.037 1.025 1.081 1.206

EC50 50.53 48.57 43.93 57.5

R2 0.998 1.000 0.998 1.000

Sig/Bkg 20.65 24.41 16.42 14.94

Table 3: **Incubation times that resulted in the best signal to noise ratio.*Original incubation time increased.

Quality control sample

Concentration (ng/mL)

Total number of samples tested

in validation

Precision –Number of

samples < 20 %CV

Relative error - number of

samples < 25 %RE

LQC1 10.0 8 8 8

MQC1 50.0 8 8 8

HQC1 125.0 8 8 8

LQC2 10.0 8 8 7

MQC2 50.0 8 8 8

HQC2 125.0 8 8 6

Accuracy and Precision Validation Data

Table : Validation data showing the accuracy and precision of quality controls in Cynomolgus monkey serum. 4

Matrix Interference Observed in Individual Serum

Serum lot number

Concentration (ng/mL)

Precision (%CV)

Back-calculated concentration

Relative error (%RE)

298057 125.0 9.5 55.89 55.3

10.0 8.4 6.79 32.1

298060 125.0 1.6 58.81 53.0

10.0 6.4 7.25 27.5

298055 125.0 3.5 79.61 36.3

10.0 1.6 5.1 48.8

Table 5: 4% Human serum spiked at the high and low quality control concentrations.

Serum lot number

Concentration (ng/mL)

Precision (%CV)

Back-calculated concentration

Relative error (%RE)

298057 125.0 3.5 138.5 10.8

10.0 3.4 9.58 4.2

298060 125.0 1.6 143.26 14.6

10.0 6.4 9.51 4.9

298055 125.0 3.5 148.0 18.4

10.0 1.6 10.14 1.5

Serum Treated with 10% DMSO

Table 6: 4% Individual lots of human serum spiked at the high and low quality control concentrations.

Matrix Effect Validation Data

Matrix Lot number LQC Met %RE ≤ 25%

%RE HQC Met %RE≤25

%RE

Human serum

298035 Yes 19.0 Yes 18.7

298037 Yes 19.7 Yes 5.1

298038 Yes 19.0 Yes 15.8

298057 No 28.6 Yes 20.2

298058 Yes 19.1 Yes 4.1

298064 No 31.4 Yes 24.9

Table 7: Minimal matrix effect observed in 2% individual human serum during validation. Serum was not treated with MgCL2 of DMSO.

Correcting Assay Drift

• Add standards, QCs and samples to non-binding plate before transferring to binding plate with a 12-channel pipette.

•*Used a 96-well ELISA plate with a higher binding effi ciency.

•Evaluated various orientation of standards and QCs.

•Evaluated various substrate incubation temperatures

* Changing the 96-well plate to a plate with higher binding e� ciency resulted in a drastic improvement of the assay drift.

Matrix Interference

• Six randomly selected individual human serum were tested for matrix interference.

• Serum were spiked and tested at the high and low quality control concentrations.

• Most individual serum had a back-calculated value that was lower than the nominal concentration and > 25% RE.

Correcting Matrix Interference

• Treat 4% human serum with 10% DMSO, 0.5M MgCL2 + 10% DMSO and 1M MgCL2 + 10% DMSO.

• Treat 4% human serum with 0.5M MgCL2, and 1M MgCL2.

• 4% individual serum lots were treated with 5% and 10% DMSO.

Matrix E� ect End Result

• Matrix eff ect was drastically improved with 10% DMSO.

• The presence of MgCL2 increased the c-value of the standard curve.

• More assays are required to complete the development of the method using DMSO and MgCL2.

• We were able to surmount the challenge of matrix interference by lowering our LLOQ and using a 1:2 dilution of human serum.

l