Taube, Sheila E.

4052 Background: 18q LOH may be a useful marker in stage II colon carcinoma to assess who needs adjuvant chemotherapy. The objective of this study was to evaluate the intra- and inter-laboratory reproducibility of an assay for 18q LOH that is similar to the integral assay used in E5202, the current Intergroup phase III stage II colon carcinoma clinical trial. Methods: Specimens from 130 stage II/III colon cancer patients were obtained from the Cooperative Human Tissue Network and evaluated at three different laboratories. Samples were divided into subsamples, coded and distributed. Each laboratory used similar equipment in an established PCR protocol with 5 dinucleotide repeats. Starting materials used both histologic sections and extracted DNA samples of tumor and non-neoplastic tissue obtained only by scalpel microdissection. LOH was defined as tumor/normal peak ratio that was >1.35 or <0.67 without microsatellite instability (MSI). Tumors were categorized as either MSI+, LOH + or -, uninformative/monoallelic, or unevaluable (DNA not amplifiable). Standard tests assessed reproducibility of both the numerical ratios and the tumor 18q LOH status. Results: Sample quality issues prevented analysis of 13 tumors (10%). The frequency of 18q LOH was 64%. The probability of agreement on the same tumor evaluated twice was 92% (Cohen's kappa 0.83). Intra-class correlation coefficients for individual markers were consistently over 0.90. Tissue specimens must have high (>85%) tumor cell content for reproducibility. Differences among patients affected reproducibility much more than effects of laboratory and sample type (DNA vs. histologic section). Conclusions: This study validates reproducibility of the 18q LOH assay for both clinical categorization of status and numerical ratios if samples are high quality and appropriate for scalpel microdissection. [Table: see text]