Improving peri-operative staging in colorectal cancer

The last few decades have seen a dramatic increase in the treatment options for colorectal cancer. These include adjuvant and neoadjuvant chemotherapy and radiotherapy, extended resections, local excision, and deferral of surgery amongst others. These developments are increasing the need for accurate staging of colorectal cancer through imaging; many treatment decisions are now taken before histopathology is available, optimal outcomes are achieved if the first surgical procedure is the definitive procedure, and histopathology results which have determined the need for adjuvant therapy are now frequently affected by neoadjuvant treatment. Assessment of imaging has predominantly judged its accuracy by comparing it to histopathology as a “gold standard”. However, this approach may underestimate the accuracy of imaging due to inaccuracies in histopathological assessment. It may also fail to focus on the adverse risk features which are most reliably identified on imaging. This thesis investigates the role of MRI in improving perioperative staging of both colon and rectal cancer. In a retrospective study of rectal cancer patients undergoing pre-treatment PET/CT and CT, I demonstrated that patients with high-risk features on rectal MRI have an odds ratio of 6 for synchronous metastatic disease compared to those with low-risk features. This suggests that, at least in rectal cancer, MRI is not only accurate when compared to histological staging, but can independently risk stratify patients with systemic disease. The next aim of this thesis was to determine whether it is possible to extend the benefits of local cancer staging with MRI seen in rectal cancer patients to patients with colon cancer. I, therefore, went on to develop a number of MRI sequences in healthy volunteers which allowed good visualisation of the colon. I then applied these to a cohort of colon cancer patients in an attempt to obtain more accurate local staging using 1.5T MRI and 3T than is currently achieved with CT. Overall, I was not able to demonstrate that MRI was significantly more accurate than CT in the local staging of colon cancer, but I was able to obtain a similar accuracy in T staging with 3T MRI (56-76%) to that achieved with CT (66-72%). In the best multiplanar 3T MRI scans, T-staging accuracy was greater than CT (67-79%). In the final part of this thesis, I investigated whether ex-vivo specimen MRI could improve that accuracy of histological staging by helping to select tumour blocks. I hypothesised that part of the “inaccuracy” demonstrated by imaging staging was due to sampling error in the selection of sections for histopathology, and that performing a high-resolution ex-vivo MRI of the specimen might reduce this error by helping to select tumour blocks. I was unable to demonstrate any benefit from this technique.