Background

There is an unmet need for real-time intraoperative colorectal tissue recognition, which would promote personalised oncologic decision making. Rapid Evaporative Ionization Mass Spectrometry (REIMS) analyses the composition of cellular lipids through the aerosol generated from electrosurgical instruments, providing a novel diagnostic platform and surgeon feedback.


Thesis Hypothesis

Colorectal lipid metabolism and cellular lipid composition are associated with the phenotype of colorectal adenomas and carcinomas, which can be leveraged for tissue recognition in vivo.


Methods

This thesis contains three work packages. First, a method for REIMS spectral quality control was developed based on a human dataset and analysis of a porcine model assessed the spectral impact of technical and environmental factors. Second, an ex vivo spectral reference database was constructed from analysis of human colorectal tissues, assessing the ability of REIMS for tissue recognition. Finally, REIMS was translated into the operating theatre, for proof-of-principle application of during transanal minimally invasive surgery (TAMIS).


Results

Sensitivity analyses revealed seven minimum quality criteria for REIMS spectra to be included in all future statistical analyses, with quality also impacted by low diathermy power, coagulation mode and tissue contamination. Based on tissue of 161 patients, REIMS could differentiate colorectal normal, adenoma and cancer tissue with 91.1% accuracy, and disease from normal with 93.5% accuracy. REIMS could risk-stratify adenomas by predicting grade of dysplasia, however not histological features of poor prognosis in cancers. 61 pertinent lipid metabolites were structurally identified. REIMS was coupled to TAMIS in seven patients. Optimisation of the workflow successfully increased signal intensity, with tissue recognition showing high accuracy in vivo and identification of a cancer-involved margin.


Discussion

This thesis demonstrates that REIMS can be optimised and applied for accurate real-time colorectal tissue recognition based on cellular lipid composition. This can be translated in vivo, with promising results during first-in-man mass spectrometry-coupled TAMIS.