A novel mechanism for accurately determining lymph node status in colorectal cancerconsensus transcriptomic profiles

Introduction: Techniques at determining lymph node positivity (prior to resection) remain poor. The ability to predict lymph node status, based on preoperative biopsies, would greatly inform our ability in relation to planning treatment for colorectal cancer. This is particularly relevant following complete pathologic response, or in polyp-detected cancers. This study aims to generate consensus profiles of dysregulated genes across numerous experiments and utilize these profiles as classifiers in differentiating nodal status. Methods: A customized graphic user interface (called CPD) was generated to develop consensus transcriptomic profiles from early and late stage colorectal cancer, based on data derived from PGER. Classifiers were generated that differentiated node positive from negative tumors. These were tested against separate Affymetrix and Illuminabased experiments. Accuracy was determined using sROC curves and a network-linkage analysis permitted development of a further consensus metastatic pathway. Results: Four consensus profiles were generated using CPD permitting development of classifiers that differentiated lymph node negative and positive colorectal cancer. Accuracy in differentiating stage two and three colorectal cancer ranged from 75 to 97 % and was both data set and classification technique-dependant. Areas under sROC curves ranged from 0.73 to 0.86. Further validation confirmed associations with the colorectal metastatic process identifying established and novel therapeutic agents. Conclusions: We have developed a transcriptomic technique that permits the determination of lymph node status from biopsy specimens. This outperforms all known radiologic approaches. This technique will enable a data-driven mechanism for planning treatment in patients with complete pathologic response or in polyp-detected cancers.