Nov 17, 2014

Applying Tissue Phenomics to Colorectal Clinical Questions

Surgical resection is considered curative for Dukes B colorectal cancer patients, however 20-30% of patients experience disease recurrence and disease specific death. We aim to stratify Dukes B patients into high and low risk subgroups through novel image based analysis algorithms. Firstly we developed an image analysis algorithm to quantify and assess the prognostic value of three histopathological features; lymphatic vessel density and invasion as well as tumour budding. Image analysis provides the ability to standardize quantification across institutes and negates observer variability. Secondly we investigated if novel histopathological features can be identified through a Tissue Phenomics approach.
Colorectal tissue sections were labelled for epithelial cells =pan cytokeratin=, lymphatic vessels =D240= and nuclei =DAPI= through immunofluorescence. The labelled tissue was used to segment and quantify tumour buds, lymphatic vessel density and lymphatic vessel invasion through automated image analysis using Definiens software. We next performed quasi-unbiased multi-parametric image analysis on the labelled tissue to quantify the complexity of the cancer microenvironment. The resultant phenome based multi-parametric signature, coupled with data mining statistics, is used to discover novel prognostic features in a Tissue Phenomics approach.