The Banff Initiative on Quality Assurance in Transplantation: Immunohistochemistry for BK Virus in the Kidney
Parmjeet S Randhawa, Samantha Chan, Jessie Climenhaga, Geng Zeng, Heinz Regele, Robert Colvin, Michael Mengel. University of Pittsburgh, Pittsburgh, PA; University of Vienna, Vienna, Austria; Harvard Medical School, Boston, MA; University of Alberta, Edmonton, Canada
Background: BK virus is an important pathogen in the kidney. Immunohistochemistry is the gold standard for diagnosis of BKV nephropathy, but its inter-laboratory reproducibility has not been evaluated.
Design: A tissue microarray(TMA) with 23 sample cores was constructed from nephrectomy specimens representing the full spectrum of BK infection from -VE to strongly +VE cases. This TMA allowed distribution of nearly identical sections to 81 participants at 60 centers. Each participating center stained the TMA using local protocols. Cases were categorized as 0, A, B or C based on staining intensity and 0, 1, 2 or 3 based on % of +ve tubular nuclei. BKV RNA was quantitated in separate samples of each tissue using real time PCR. Details of local staining protocols and evaluation scores of each participant were collected via an online survey. Stained slides were returned for centralized panel scoring. Weighted kappa (K) statistics were used to determine the inter-observer variability (comparison of local reads to panel scores) and inter-laboratory variability (comparison of panel reads on TMAs prepared at different centers).
Results: Participating centers generated a total of 2025 data points for analysis. Inter-observer reproducibility was substantial (mean kappa score (κ) = 0.64) However, using a reference case with the brightest staining, inter-laboratory reproducibility of the intensity of the stain and % nuclei was below chance (mean κ = -0.22), though it improved if only agreement on the % tubular staining was considered (κ = 0.29). Reproducibility for staining intensity remained poor, but improved (κ = 0.22) if it was based on the consensus reading for each case. Binary categorization of each biopsy as BKV positive or negative resulted in the highest agreement (mean κ = 0.77). BKV DNA content per cell increased progressively with increasing intensity of immunohistochemical staining (median BKV DNA per cell: 4.88E+01 in grade0, 5.82E+01 in grade A, 2.92E+04 in grade B, and 4.60E+04 in grade C). The threshold sensitivity of detection of immunohistochemistry was estimated to be 58.2 BKV genomic equivalents per cell.
Conclusions: There is good agreement between observers on the interpretation of a BKV stain, particularly if only the presence or absence of BKV is assessed. However, quantitative scoring of nuclear staining intensity, and percentage tubules infected as proposed in the Banff 2009 Conference on Allograft Pathology is not reproducible.
Category: Kidney (does not include tumors)
Tuesday, March 20, 2012 9:00 AM
Platform Session: Section H, Tuesday Morning