M1078. Case Study: Comprehensive Incurred Sample Reanalysis Investigation
for Clinical Sample Analysis
Nadezhda Kulagina, John Kamerud and Joseph Bower
Immunochemistry Services, Covance Laboratories Inc., Chantilly, Virginia
Introduction
ISR Investigation Results
Sample Relationship
Table 2. ISR Investigation for Analyte in Human Serum:
Comparison of Original and Repeat Results
The performance of a validated method employing ligand-binding
assay is monitored during samples analysis using appropriate quality
control (QC) samples that are typically prepared by spiking the
analyte in the matrix of the study samples. Although performance of
the spiked QC samples provides a reasonably good measure of
method reproducibility, there may still be uncertainty about the
reproducibility of results generated for study samples (incurred
samples) in which the analyte may have been subjected to a more
complex biological system. The February 2008 AAPS Workshop
provided recommendations related to the conduct and expectations
of the incurred sample reanalysis (ISR) in order to demonstrate
reproducibility of bioanalytical or immunochemistry methods for
quantification of both small and large molecule drug entities.
Figure 3. Evaluation of failed ISR samples per subject and cohort
Purpose
■
To discuss incurred sample reanalysis (ISR) investigation for failed
ISR samples
■
To evaluate impact on clinical sample analysis
Method
■
■
Sandwich bioassay with electrochemiluminescence (ECL)
detection was used in the study (Figure 1)
An in-study assessment of the analytical method was performed
through reanalysis of a sub-set of samples. The criteria for
reanalysis included but was not limited to:
• At least 2/3 of reanalyzed samples should be within ± 30 of
the original value
• Samples (106) selected for reanalysis samples had original
concentration values between the Low and High QC values
Figure 2. Comparison of original and ISR values.
■
ISR samples demonstrated a relative over recovery as compared to
the initial result (11 of 106 ISR samples were lower than the initial
and 95 ISR samples were greater than the original result).
■
ISR Plates
ISR samples were analyzed on plates ISR-1, ISR-2, ISR-3, and ISR4. All ISR plates contained samples which demonstrated >30%
difference from the initial result indicating that the ISR plate was not
a primary factor in out-of-range performance.
Samples from Caucasian (C) and Asian (A) subjects (3 cohorts) included in ISR. Each subject
was represented by at least 6 samples. Asian and Caucasian subjects had a total of 41% and
21 % of ISR samples that exceeded 30% difference, respectively. Out-of-range samples
appeared to be grouped by subject (i.e.º a subject has several samples >30% difference or a
subject has no samples out-of-range). The relationship of subject or cohort specific to ISR
cannot be excluded. This performance difference could potentially be related to selectivity
which could be impacted by a change in coating reagent lot.
Variability between the original and investigation repeat results with respect to original result,
determined by two analysts using both lot A and B coating materials, was within 30%
acceptance for the majority of results.
Table 3. ISR Investigation for Analyte in Human Serum:
Comparison of ISR and Repeat Results
Twenty samples from the original ISR analysis were assayed again by analysts I and II that
contributed to original and ISR data, respectively. Each analyst’s plate contained the same 20
samples of which 10 samples demonstrated >30% difference from original in the ISR
evaluation and 10 samples were within 30% difference from original.
Table 1. ISR Investigation for Analyte in Human Serum:
Analyst Variability
Various Parameters
There were no apparent association of using different equipment,
lots of Calibrators and QCs, lots of detection reagent and sample
freeze-thaw cycle (analyzed within 2–3 cycles) in ISR out-of-range
performance.
Analyst
All ISR plates were completed by analyst II while all initial results
were by analysts I and III. Due to the general over recovery of ISR
relative to initial results, the impact of an analyst cannot be excluded.
Figure 1. Schematics of sandwich assay for detection of analyte.
ISR Investigation Plan
■
> 1/3 of the results (38 of 106) were outside 30% difference of
original value
■
Initiation of comprehensive ISR investigation. The following
variables were evaluated: ISR runs, calibrator and QC
performance, sample freeze/thaw, coating and detection reagents,
equipment, analyst performance variability, and sample relationship
Variability between the ISR and investigation repeat results with respect to ISR result,
determined by two analysts using both lot A and B coating materials, showed higher variability
especially for previously failed samples suggesting a possible error during ISR run.
Coating Material
Lot A of coating material was used for all initial results and lot B was
used for all ISR plates. Standard reagent qualification was
successfully applied prior to changing coating reagent lots however
the impact of coating reagent lot cannot be excluded. This reagent
difference has the potential to impact selectivity.
Conclusions
Investigation of sample repeat showed that although all results were within 30% difference of
analyst I value, higher result variability (8.7% to 26.2%) was observed between analysts I and II
on plates coated with lot A (used to generate original results) coating material. In addition, higher
result variability (10.2 to 33.1%) was observed between analyst’s II data generated on plates
coated with A and B materials.
■
Despite a thorough investigation for failed ISR no assignable cause was identified. The repeat analysis
suggested no apparent association of ISR failure with the use of different lots of coating materials or other
assay reagents. Repeat analysis of failed samples demonstrated agreement with the original results, as well
as no analyst bias.
■
While the composition of the study population did show some evidence for potential bias, this could not be
concluded from the data. Based on the results it was concluded that original results are valid and the
method could be used to support a clinical study. Later the method reproducibility was confirmed by ISR
results in two other independent clinical studies.
M1078
Case Study: Comprehensive Incurred
Sample Reanalysis Investigation for
Clinical Sample Analysis
Nadezhda Kulagina, John Kamerud and Joseph Bower
Immunochemistry Services,
Covance Laboratories Inc., Chantilly, Virginia
Presented at the
2011 AAPS National Biotechnology Conference
San Francisco, California
16–18 May 2011
Covance is an independent, publicly held company with headquarters in Princeton, New Jersey, USA.
Covance is the marketing name for Covance Inc. and its subsidiaries around the world.
The Americas +1.888.COVANCE ( +1.888.268.2623)
+1.609.419.2240
Europe/Africa +800.2682.2682 +44.1423.500888
Asia Pacific
+800.6568.3000 +65.6.5677333
Web Site: www.covance.com
© COPYRIGHT 2011, COVANCE INC.