This document summarizes a study that used quantitative image analysis of HER2 expression in esophagogastric adenocarcinoma tissue samples to identify prognostic factors. Automated image analysis was performed using the Definiens platform to extract over 50 image features from segmented cells and subcellular compartments. Multivariate regression analysis was able to predict disease-free and overall survival times from these quantitative image features, which were then shown to have prognostic value when analyzed with Kaplan-Meier survival curves. This demonstrated that automated quantitative image analysis can provide statistically significant prognostic factors not accessible to visual analysis alone.
1390 Identification Of Prognostic Factors Using Quantitative Image Analysis Of Her2 Expression.Pdf 1390
1. Identification of Prognostic Factors using
Quantitative Image Analysis of HER2 Expression
by Immunohistochemistry (IHC) in
Adenocarcinoma of the Esophagogastric Junction
Günter Schmidt, Gerd Binnig
Definiens AG München
Annette Feuchtinger, Axel Walch
Pathology, HelmholtzZentrum München
52nd Symposium of the Society for Histochemistry
Prague, 1 - 4 September 2010
2. Study Overview
Surgical Resection Prognostic factor performance
Klinikum Rechts der Isar, TU Munich Definiens AG; Biomathematics and
Biometry, Helmholtz Zentrum
Visual HER2 scoring by pathologist
Pathology, Helmholtz Zentrum
Illustration
Image: University of California, 1919
Tissue IHC staining and
image acquisition
Pathology, Helmholtz Zentrum Definiens Developer XD, 2010
Slide - 2 Quantitative image analysis
Definiens AG
3. Data: Tissue Micro Arrays of Biopsy Tissue Sections
� 132 cancer patients
� 390 tissue cores on 3 TMAs
� HER2 (human epidermal
growth factor receptor 2)
� Membrane protein
� Known to indicate
aggressive cancer subtypes
Slide - 3
4. Pathologist Score 3+
Score depends an membrane staining intensity, staining completeness,
and percentage of stained tumor cells
5x
20x
Slide - 4