Algorithms for quantitation of protein expression variation in normal versus tumor tissue as a prognostic factor in cancer: Met oncogene expression, and breast cancer as a model

Abstract

Immunohistochemistry and immunofluorescence (IF) assays frequently rely on subjective observer evaluation for grading. The aim of our study was to develop an objective quantitative index based on confocal laser scanning microscopy (CLSM) and image analysis of an IF assay to determine alteration in protein expression levels in normal versus tumor tissue. The relative levels of Met expression, a prognostic factor in breast cancer, were used as a model for evaluating image analysis algorithms. Primary human breast cancer biopsies were collected. Sections containing tumor and adjacent uninvolved normal regions were immunostained for Met and digital images were acquired by CLSM. Subsequently, the digital data were manipulated using several different algorithms to calculate prognostic indexes. The results were correlated with the clinical outcome to determine the prognostic value of these indexes. Different algorithms were used to obtain quantitative indexes to evaluate the relative levels of Met expression. We report a statistical correlation between patient prognosis and relative Met level in normal versus tumor tissue as determined by three distinct algorithms using Kaplan-Meier analysis (log-rank): calculations based on intensity levels differences DV (P = 0.002), DIntensity (P = 0.014), and entropy divergence (Dentropy; P = 0.0023). Using adjacent normal tissue as an internal reference, a quantitative index of tumor Met level divergence can be objectively determined to have a prognostic value. Moreover, this methodology can be used for other proteins in a variety of different diseases.