Grant Number:	5R03CA077125-02 Interpret this number
Primary Investigator:	Glaser, Sally
Organization:	Cancer Prevention Instit Of California
Project Title:	Reliability of EBV Detection in Hodgkins Disease Tumors
Fiscal Year:	1999

DESCRIPTION (Applicant's Description) This proposed R03 study will be the first to examine observer reliability of EBER in situ hybridization (ISH) and LMP-1 immunohistochemistry (IHC), morphologic molecular assays widely used in clinical studies to detect Epstein-Barr virus (EBV) latent infection of tumor cells. As these assays are easy to use, rapid, and applicable to archived tumor specimens, they are ideally suited for large-scale molecular epidemiologic projects. In Hodgkin's disease (HD), a common cancer of young adults, EBER ISH and LMP-1 IHC have detected EBV in about 40 percent of cases, and marked epidemiologic differences exist between the virus-defined subgroups, However, EBV prevalence varies among HD studies even after statistical control for epidemiologic differences in populations. Problems with observer reliability may be partly responsible; yet, although assay interpretation is subjective, reliability has never been evaluated for either test. This project will determine inter- and intra-rater reliability of EBER ISH and LMP-1 IHC assays in EBV detection in 40 HD tumors, based on two reviews by each of four hematopathologists of slides, prepared at each of two labs. From tumor specimens evaluated for prior projects, we will randomly select 10 EBV-positive and 10 EBV-negative cases of each of the two most common histology subtypes, to maximize precision of the kappa statistic. From these 40 banked specimens, each lab will stain 40 slides for EBER ISH and 40 for LMP-1 IHC, yielding 80 slides per assay. At NCCC, stained slides from the two labs will be intermixed within assays, relabeled to blind raters to institutional identification, and mailed in batches to raters, who will record their interpretations (EBV-positive, EBV-negative) and both rank and describe interpretive difficulty. Been lst and 2nd reviews, staff will renumber slides to blind raters to the review sequence. 640 ratings (320 first and 320 second) will occur for each assay. With these data, we will compute kappa statistics for multiple raters to determine inter- and intra-rater agreement occurring beyond chance for each assay. Kappas will also be compared for agreement on slides prepared on the same specimens by each lab to detect any lab effect on interpretation accuracy, and for the two histologic subtypes to detect differences that may lead to variation in EBV prevalence by subtype. Findings will be developed into specific interpretation guidelines. Study strengths include: the first examination of both inter- and intra-rater agreement for two commonly used clinical EBV detection assays with potential for epidemiologic application in studying EBV-linked cancer; development of explicit interpretation guidelines; use of existing pathology materials for efficiency and for assembling a case series which maximizes precision; adequate sample size to estimate inter-histology differences with sufficient statistical power; collaboration with experts in the field of EBV detection; and focus on HD, a common EBV-related cancer with both epidemiologic and clinical importance for EBV classification.





None