Detection and staging of colonic lesions using computed tomography and magnetic resonance imaging

Abstract

In Sweden, more than 6000 new patients were diagnosed with colorectal cancer in 2015 of which over 4000 patients had colon cancer and 1800 died from the disease. It is the second most common cancer after breast- and prostate cancer. In the last decade, significant improvement in the treatment of both rectal and colon cancer have been achieved. Diagnostic imaging, using CT, MRI and PET/CT, has become essential in the preoperative work-up. New neoadjuvant treatment strategies are under study in colon cancer. In the selection of patients for these treatments, pre- and post-treatment imaging has also become of interest. The overall aim of this thesis is to evaluate cross sectional imaging modalities for detection of colonic polyps and staging of patients with colon cancer using CT and MRI.

The aim of paper 1 was to investigate the impact of radiation dose and spatial resolution in detecting colonic polyps in a phantom study simulating computed tomographic colonography (CTC). By using different scanning protocols with different slice -thickness, pitch and tube current we showed that the dose level could be substantially reduced by lowering the tube current without compromising the detection rate for polyps larger than 5 mm.

The aim of paper 2 was to evaluate if high resolution MRI of colon cancer contributed to the standard staging procedure with CT with respect to assessment of local tumour extent, nodal staging and extramural venous invasion (EMVI). An advantage of MRI over CT due to its soft tissue discrimination to identify prognostic factors such as tumour stage and extramural venous invasion was found. The result of nodal staging for both modalities were equally moderate.

The aim of paper 3 was to evaluate commonly used imaging CT criteria for lymph node metastases in predicting stage III disease. Of the different imaging criteria, morphological features performed best specifically internal heterogeneity and irregular outer border. None of the size criteria were predictive.

The aim of paper 4 was to validate morphological CT criteria from paper 3 in a prospectively collected patient cohort using two observers. By using the criteria internal heterogeneity and a combination including irregular outer border, a moderate sensitivity and high specificity was achieved predicting stage III disease.

CT and high resolution MRI can be used to classify colonic tumours into not locally advanced or locally advanced. The prediction of lymph node metastases with the most commonly used image modalities is however unsettled and challenged. In the setting of selecting patients to neoadjuvant chemotherapy, the ongoing trials so far have used inclusion criteria based on tumour T-stage only (T3cd-T4 as locally advanced). Patients with lower tumour T-stage but still have other adverse prognostic feature such as regional metastases will therefore potentially be undertreated and patients with no metastases will potentially be overtreated. Search for other prognostic factors identified on cross sectional imaging has to be performed.