[2139] Novel Quantitative Image-Analysis Based Scoring Technique for In-Situ Assessment of mRNA in Archival Tumor Tissues: Strong Correlation between Manual and Automated Schemes

Jeffery C Hanson, Timothy R Holzer, Angie D Fulford, Robert J Konrad, Aejaz Nasir. Eli Lilly & Co., Indianapolis; Laboratory for Exp Medicine, Indianapolis, IN

Background: We developed a semi-quantitative scoring technique for manual in-situ mRNA quantification of oncology biomarkers in archival tumor tissues. We also developed an image analysis-based quantitative algorithm as an automated in-situ mRNA evaluation tool. The goal was comparative evaluation of these scoring techniques for novel oncology biomarkers.
Design: Archival sections from 6 non-small-cell lung carcinomas (NSCLCA) were stained for BIRC5 (survivin) mRNA. The in-situ mRNA staining was manually scored by an experienced pathologist as 0, 1+/2+ (<50%/≥50% area of the cell with non-confluent fine, red granules) or 3+/4+ (with partial/total confluence of granules). The sum of weighted mRNA scores (staining scores X respective %ages) was reported as a semi-quantitative (sq-) m-RNA score. Whole slide images were captured (Aperio Scanscope XT) and analyzed by Definiens' EII software. Each tumor section was scored using the manual criteria above and a quantitative (q-) m-RNA score was derived. Both scores (range 0-400) were correlated (Pearson test; GraphPad, Prism).

Results: Manual sq-RNA scores on 6 NSCLCA cases ranged 0-162. The automated algorithm analyzed tens of thousands of cells/section. The positive cells ranged from 2.5% to 75%, and the q-RNA score ranged from 2-120. Key challenges with the automated technique were delineation of various tissue/tumor components and tissue artifacts. The image analysis-based q-RNA scores correlated welll with the manual scores (Pearson's correlation coefficient of 0.967; p=0.002).

Conclusions: Our quantitative image analysis-based in-situ mRNA scoring on archival tumor tissues shows high-correlation with the manual techniqe. Evaluation on larger cohorts of independent test sets of archival tissues will further substantiate its practical utility in oncology clinical trials and practice.
Category: Techniques

Monday, March 19, 2012 1:00 PM

Poster Session II # 308, Monday Afternoon


Close Window