Reproducibility for C4d immunohistochemistry in renal allografts

1
Reproducibility for C4d immunohistochemistry in
renal allografts results from the Banff Trial

• Banff Initiative for Quality Assurance in
  Transplantation (BIFQUIT)
• Samantha Chan, Jessie Climenhaga, Parmjeet
  Randhawa, Heinz Regele, Yaël Kushner, Robert
  Colvin,
• and Michael Mengel
• University of Alberta, Edmonton, Canada
• University of Pittsburgh, Pittsburgh, USA
• Massachusetts General Hospital, Boston, USA

2
Detection of C4d is crucial for diagnosing
antibody mediated rejection reproducibility
studies are limited
C4d in paraffin sections Results can directly
influence patient care
3
Design of the C4d BIFQUIT trial
Collect cases (n19) for analytical spectrum
(i.e. Negative, mildly to strongly positive)
Sent slides requested by participating
institutions
Participants/ observers stain slides and evaluate
Tissue microarrays (TMA)
Centers contributing samples Volker Nickeleit,
Chapel Hill, USA Verena Bröcker, Hannover,
Germany Parmjeet Randhawa, Pittsburgh, USA
Bernard Collins, Boston, USA Heinz Regele,
Vienna, Austria Michael Mengel, Edmonton,
Canada Cinthia Beskow-Drachenberg, Maryland,
USA Surya Seshan, New York, USA
Complete online or paper survey on staining
methods C4d scoring
Mail survey stained slides back to organizing
centre
Combine with retrieved online survey data
Enter survey data into data base
ALL TMA slides for score by review panel the
best stain returned was identified per consensus
as the reference case
Data Analysis
Collect and organize slides (e.g. by
institution/ participant)
4
Reproducibility of immunohistochemical stains

• Inter-institutional variability
• staining/laboratory procedure
  subjectivity/interpretation by observer
• weighted kappa values were calculated by
  comparing the C4d scores provided by each
  participant from their locally stained slides to
  the corresponding C4d scores provided by the
  local participant from the reference case
• Inter-laboratory variability
• staining/laboratory procedure
• weighted kappa values were calculated by
  comparing the panel consensus read for each
  tissue core on each slide to the panel read of
  the corresponding tissue core on the reference
  slide
• Inter-observer variability
• subjectivity/interpretation by observer
• weighted kappa values were calculated by
  comparing the reads of the local participants to
  the consensus reads of the panel recorded on the
  same TMA slide

5
Summary of mean kappa-values for the different
components influencing the reproducibility of a
C4d stain
mean Kappa-value comparing Banff C4d scores to reference case mean Kappa-value comparing Banff C4d scores to majority calls by all participants mean Kappa-value comparing positive/negative C4d calls to reference case
Inter-institutional reproducibility 0.17 0.3 (-0.68 0.65) 0.26 0.2 (-0.5 0.64) 0.65 0.3 (-0.46 1.0)
Inter-laboratory reproducibility 0.46 0.6 (-0.77 1.0) 0.60 0.4 (-0.36 1.0) 0.78 0.4 (-0.66 1.0)
Inter-observer reproducibility 0.45 0.2 (0.11 0.9) NA 0.60 0.4 (-1.0 1.0)
Significance of kappa values lt0 indicating no
agreement (or less agreement than expected by
chance), 00.20 slight, 0.210.40 fair, 0.410.60
moderate, 0.610.80 substantial, and 0.811
almost perfect agreement. NA Not Applicable,
because for the inter-observer comparison per
definition the comparator has to be the panel
read and cannot be a majority call
6
Scatter plots showing the inter-laboratory and
inter-observer reproducibility for each
individual participant / participating laboratory
using the Banff C4d schema (A) or positive vs.
negative calls (B)
A using the Banff C4d grading schema
B using positive negative calls
18
59
kappa values for inter-observer reproducibility
kappa values for inter-laboratory reproducibility
kappa gt0.6 at least moderate reproducibility
kappa gt0.4 at least fair reproducibility

• Variance in scoring between observers
• was greater with C4d1 and C4d2 cases compared to
  C4d0 and C4d3 cases
• the panel consistently under-scored the local
  observers, i.e. Pathologists adjust their
  scoring to the sensitivity of their local
  laboratory

7
Influence of analytical variables on C4d
staining comparing the top 15 and bottom 15
slides ranked by kappa values for
inter-laboratory reproducibility
8
Influence of Fixation for C4d Inter-Laboratory
reproducibility
Mean kappa values
Protocol Description
1 Standard Immediate onset of fixation for 24h in buffered, standard Formalin (4)
2 Delay Delayed onset of fixation for 12h (storing the tissue at 4C) and then fixation for 24h in buffered, standard Formalin (4)
3 Frozen Snap freezing of tissue (like simulating a frozen section procedure) and then immediate fixation for 24h in buffered, standard Formalin (4)
4 Overfix Immediate onset of fixation for 5 days (simulation of shipment over long weekend) in buffered, standard Formalin (4)
5 Underfix Immediate onset of fixation for only 1h (simulation of very rush processing) in buffered, standard Formalin (4)
6 Etha Immediate onset of fixation for 24h in Ethanol (100)
9
Summary and recommendations

• The BIFQUIT trial results indicated that C4d
  staining on paraffin sections is of limited
  reproducibility.
• Refinement of the current Banff C4d scoring
  schema and standardization of tissue processing
  and staining protocols is necessary to achieve
  acceptable reproducibility for C4d staining on
  paraffin-embedded material
• Recommendations from this first BIFQUIT trial for
  C4d immunohistochemistry on paraffin sections
• Avoid under (lt1 hour) and/or ethanol fixation of
  tissue specimen
• Use heat induced epitope retrieval with citrate
  buffer at pH6-7
• Incubate polyclonal anti-C4d antibody
  concentrated at lt180 for gt40 minutes
• Better results were observed with harsher
  pre-treatment (e.g. autoclave epitope retrieval),
  higher polyclonal anti-C4d antibody
  concentrations (110), and longer incubation
  times for the polyclonal anti-C4d antibody (over
  night / 12-16 hours)
• Simplifying the 2007 Banff C4d grading schema
  will reduce inter-observer variability

10
Acknowledgements
Many thanks to participants in the Banff Working
Group Trials!
The following Institutions contributed cases to
the BIFQUIT trial Cinthia Beskow-Drachenberg,
Maryland, USA Volker Nickeleit, Chapel Hill,
USA Verena Bröcker, Hannover, Germany Bernard
Collins, Boston, USA Heinz Regele, Vienna,
Austria Surya Seshan, New York, USA
The panel
Students helped with logistics and
analysis Samantha Chan Jessie Climenhaga Victoria
Sheldon Akshatha Raghuveer
Astellas Pharma Canada provided an unrestricted
research grant to the Banff Working Groups