Page 58 - ICCS 2011 Program Web

This is a SEO version of ICCS 2011 Program Web. Click here to view full version

« Previous Page Table of Contents Next Page »
56
ICCS 2011
P20
Validation of an assay to detect paroxysmal
nocturnal hematuria (PNH) cell populations with
0.01% limit of detection: Practical and statistical
considerations
Deborah A. Katanik
1
, Joshua Coleman
1
, James F.
Bena
3
, Jaroslaw P. Maciejewski
2
, Eric D. Hsi
1
1
Department of Clinical Pathology, Cleveland Clinic,
Cleveland, OH, USA,
2
Taussig Cancer Center,
Cleveland Clinic, Cleveland, OH, USA,
3
Quantitative
Health Sciences, Cleveland Clinic, Cleveland, OH,
USA
Introduction: Recent guidelines for PNH Flow
cytometry (FC) assays recommend a lower limit of
detection (LOD) of 0.01% PNH cells. Performance of
such an assay is problematic for many laboratories
due to rarity of PNH samples and lack of familiarity
with validation requirements for rare event analyses.
Methods: We developed a FC assay to detect
granulocytes via successive doublet discrimination,
CD45-light scatter and CD15/CD33 gating.
Erythrocytes were gated using doublet-discrimination
and Glycophorin-A/light scatter. Using a statistical
model to validate a 0.01% LOD with 95% confdence,
we created a LOD validation sample set from 10
known PNH patients that consisted of 34 separate
samples (3-5 replicates/patient) diluted to 0.01%
PNH cells using normal blood diluent. PNH-type
granulocytes were defned by lack of both FLAER and
CD24 reactivity while PNH-type erythrocytes were
defned by lack of CD59. Results: Collecting suffcient
gated events (157,000) proved problematic, and
thus a lower event threshold was used in practice.
95% confdence of a positive result at the 0.01%
LOD level was evaluated for each sample using the
number of gated events. The number of samples
required was based on power analysis. Of the 34
samples with between 75,000 and 157,000 events,
33 (97%) met the threshold for positivity at 0.01%
LOD, thus validating the sensitivity of the assay.
Conclusion: High sensitivity assays are capable of
detecting 0.01% PNH cells, but our validation required
biostatistical support to validate LOD claims and easy
access to PNH samples.
POSTER ABSTRACTS