Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
© Ortho-Clinical Diagnostics, Inc.
lot to lot variation in Clinical Chemistry and Immunoassay
Norwegian Clinical Chemistry EQA Program
12 March 2015
HOULLE Jean-Pierre
Scientific Specialist International
Ortho Clinical Diagnostic European Center
STRASBOURG
© Ortho-Clinical Diagnostics, Inc.
VITROS Analyzer MICROSLIDE
System Accuracy OverviewPerformance Assignment & Release
3/27/2015 2Ortho-Clinical
The OCD Release Laboratories:Support Fluids and Slide Manufacturing for
VITROS 250, 350, 950, DT60, Fusion 5,1, 3600, 4600, 5600 Analyzer
Systems
Ortho-Clinical Diagnostics – VITROS Analyzer System Accuracy Overview
• Sample Acquisition - Obtains Patient Samples for all test events.
• Reference Laboratory – Performs Reference Method Analysis.
• Accuracy Based Release Program – Set accuracy for Slide Master Lots.
• GEN Disposition/Assignment - Releases new lots of slides, CalDisk/ADD.
• Fluids Release – Releases the calibrator and control materials.
• Control Value Assignment – Sets the acceptable Quality Control Limits.
• Microbiology – Tests for any microbial contamination in Fluids or Reagents.
• Stability Laboratory - Confirms all products meet stability claims.
• Analyzer Maintenance – Assures instrument performance.
VITROS Analyzer MICROSLIDE
3
Accuracy Based Release Program (ABRP)
VITROS Traceable Link to Accuracy
• Goal: Enhance the verification
and setting of new coating
accuracy.
Reference-based
Master slide Lot
Accuracy Based Release Program
GEN Disposition/Value Assignment Tests
Reference
Method New Coating
Ortho-Clinical Diagnostics – VITROS Analyzer System Accuracy Overview4
Define Coating CharacteristicsUsing Statistical Test Designs (Slide Precision/Quality Control)
Slit 1, 2, 3 …..
Roll x
Roll 1
Roll 3
Roll 2
Roll x
Test
Uniformity
Width-wise
Length-wise
Ortho-Clinical Diagnostics – VITROS Analyzer System Accuracy Overview5
Release Multiple Customer LotsGEN/Value Assignment Process
• Each Coating is Compared to the
Master Lot (Check Web).New Coating (Same GEN) vs Check Web
-10.00
-8.00
-6.00
-4.00
-2.00
0.00
2.00
4.00
6.00
8.00
10.00
0.00 20.00 40.00 60.00 80.00 100.00 120.00
Conc [mg/dL]
Bia
s [
mg
/dL
] C
he
ck -
Re
fere
nce
MCL Limit
New Coating vs Check Web
New GEN Required
-10.00
-5.00
0.00
5.00
10.00
0.00 20.00 40.00 60.00 80.00 100.00 120.00
Conc [mg/dL]
Bia
s [
mg
/dL
] C
he
ck -
Re
fere
nce
MCL Limit
New Coating vs Check Web
New GEN Required after Value Assignment
-10
-5
0
5
10
0.00 20.00 40.00 60.00 80.00 100.00 120.00
Conc [mg/dL]
Bia
s [m
g/d
L]
Ch
eck -
Re
fere
nce
MCL Limit
New GEN
Required
Same GEN After Value
Assignment
6
Chemistry-specific Manufacturing Control Limits (MCL) are
used to assure consistent:
Length-wise and Width-wise Uniformity
Precision
Accuracy
Value Assignment through Patient Calibrations
ID
Vitros
Response
Reference
Value
Apply Patient
Cal
Regression to
Response
Conc. = 20*
Response + 1
Patient 1 0.1 3 3
Patient 2 0.2 5 5
Patient 3 0.3 7 7
Patient 4 0.4 9 9
Patient 5 0.5 11 11
Calibrator 1 0.15 Unknown 4
Calibrator 2 0.35 Unknown 8
Calibrator 3 0.45 Unknown 10
OCDUse Reference Value as Standards to predict Calibrators
Customer
Vitros
Response
Value
Assigned
Conc.
Apply
Calibrator
Regression to
Response
Conc. = 20*
Response + 1
Calibrator 1 0.15 4 4
Calibrator 2 0.35 8 8
Calibrator 3 0.45 10 10
Patient A 0.1 Unknown 3
Patient B 0.23 Unknown 5.6
Patient C 0.34 Unknown 7.8
Patient D 0.38 Unknown 8.6
Patient E 0.47 Unknown 10.4
Customer Use VITROS Cal Value as Standards to predict Patients
1. Reference Values
used on Patient
Samples to Create a
Standard Curve
4. VITROS is Calibrated
using Calibrator Values
Assigned from Patient
Calibration
ExampleExample
2. Using Patient Cal
Curve, determine
Calibrator Values
5. Patient
Results reported.
Ortho-Clinical Diagnostics – VITROS Analyzer System Accuracy Overview7
Relate Analyzer Response of Patients to Concentrations Determined by Reference Methods
Value Assignment of VITROS Calibratorsusing a Patient Calibration
Master Lot Concentration
VITROS Response
DR
Using Patients as “Knows”;
predict the Calibrator Fluids
SAVs
Ortho-Clinical Diagnostics – VITROS Analyzer System Accuracy Overview8
10© Ortho-Clinical Diagnostics, Inc.
ERROR BUDGET
This is the tool that was
originally used to evaluate
system variability.From
the error budget release
limits were derived
11© Ortho-Clinical Diagnostics, Inc.
ERROR BUDGET
A typical error budget
contained the following
parts:
12© Ortho-Clinical Diagnostics, Inc.
Maximum Allowable Displacement
The WFI (coating to
coating) is derived from
the maximum allowable
displacement and broken
down into five
components:
13© Ortho-Clinical Diagnostics, Inc.
Performance Verifiers Level 1
14© Ortho-Clinical Diagnostics, Inc.
Performance Verifiers Level 2
15© Ortho-Clinical Diagnostics, Inc.
Example of Ca patient means
© Ortho-Clinical Diagnostics, Inc.
how we manage lot to lot bias
3/27/2015 16Ortho-Clinical
17© Ortho-Clinical Diagnostics, Inc.
Vitros ECi TSH - MGCC lot 32
QC Mean: Assigned mean from calibration
MGCC (UL/LL):
MGCC limit calculated using QC curves. Input: Maximum allowable bias, n = 20 and
risk to customer of 10%
Max Bias (UL/LL):
Product specific maximum allowable bias (total mean bias)
Mean (n=4):
Single stability time point limit for valid time point (determined using the specific
error budget and n = 4)
IGCC (n=1):
Individual assay limit for all assay replicates
B
C
D
Maximum permitted positive bias lot to lot from the calibrated control mean taking into consideration test error at MGCC test
Reduction from Maximum allowable (Max Bias) bias using Operating Curves (n=20) to set MGCC limits
Ignore C
Maximum allowable single assay result (n=1). Data remains within the statistically accepted error.
A
10
15
20
25
30
35
40
45n
g/m
L V
itam
in D
QC Mean
MGCC UL
Max Bias UL
Mean (n=4)
IGCC (n=1)
MGCC LL
Max Bias LL
Mean (n=4)
IGCC (n=1)UL = upper limitLL = lower limit
B CD
A
Trend Warning Limits for a stability study. They are typically set to the same value as the MGCC limits (tightest limit).Performance that remains within these limits supports acceptable stability over shelf life. Exceeding these limits at a single time point requires additional review.
•Individual control results should meet the IGCC (n=1) limit to assure validity of each individual replicate and, therefore, an individual assay.•Overall mean performance should remain within the Max Bias Limits.
Slide no 19
Slope of the calibration curve (dose %CV/signal %CV)
Variation in instrument signal output within-run
Variation between well signal output
Variation in signal output due to SR (age and lot) and WR
Variation in instrument signal output between days
Current error budget structure
Variation in dose due to reading from different cal curves
Variation in dose due to using different lots of reagent
Variation in dose due to differences between labs and instruments
Slope factor
Within Machine
Well to well
Within day
Generic Reagents
Day to day variation
Within Calibration
Curve Fit (Calibration)
Within Lot
Batch to Batch
Within Lab
Lab to Lab
Total Variation3/27/2015
20© Ortho-Clinical Diagnostics, Inc.
Vitros Eci- Calibration Quality parameters
Differences in luminometer response between
instruments, variables from the Spread Limit and signal
changes over the shelf-life of the reagents can affect the
signal of the calibrator
Example of the contribution for the TT3 assay:
Instrument to instrument 20%
Within instrument 13%
Age of the reagents - 30%
Total + 33 to - 63%
21© Ortho-Clinical Diagnostics, Inc.
Vitros ECi- Calibration Quality Parameters
The error budget is the maximum percentage change by
which different areas of the system can vary
Example of the error budget for the TT3 assay:
Sample Metering 2.1%
Reagent Metering 1.0%
SR Metering 1.0%
SR batch to batch 3.0%
Well to well 4.0%
Incubator 0.9%
Luminometer 1.0%
Total 13%
Slide no 22
What do we use them for?Setting curve shape limits
Setting well to well limits
Setting instrument release limits
Deriving limits based on securing precisionSetting single run limits
Setting MGCC limits
Setting selling control ranges
Slope factor
Within Machine
Well to well
Within day
Generic Reagents
Day to day variation
Within Calibration
Curve Fit (Calibration)
Within Lot
Batch to Batch
Within Lab
Lab to Lab
Total Variation3/27/2015
Defined using clinical need as an input
Slide no 23
MGCC/IGCC tests
• Product performance limits– Mean control bias at MGCC
– Master curve shape at MGCC
– Limits at IGCC
• Test performance limits– Individual run limits (controls and curve shape)
– Control precision limits at MGCC
3/27/2015
24© Ortho-Clinical Diagnostics, Inc.
Manufacturing internal QC controls
25© Ortho-Clinical Diagnostics, Inc.
Manufacturing internal QC controls
26© Ortho-Clinical Diagnostics, Inc.
Troponin I:Typical lot to lot variation observed on Cliniqa controls
mean sd %CV
reagent lots 1771 1780 1790 1800
level 1 0.087 0.091 0.085 0.085 0.087 0.003 3.25
level 2 0.614 0.616 0.593 0.576 0.600 0.019 3.16
level 3 6.66 6.67 6.52 6.36 6.55 0.15 2.22
27© Ortho-Clinical Diagnostics, Inc.
Moving Average
30© Ortho-Clinical Diagnostics, Inc.
31© Ortho-Clinical Diagnostics, Inc.