Introduction MRI has proven extremely useful in pre-operative staging of rectal cancer. It has replaced CT scanning in assessment of the pelvis to become the current basis for Multi-Disciplinary Team discussion and decision making in relation to neoadjuvant therapy. However, recent studies correlating MRI and histopathology results for all rectal cancers have found a poorer performance for MRI in low rectal cancers. This study aimed to assess and compare the performance of CT and MRI scanning in assessing lymph node positivity in rectal cancers requiring abdomino-perineal excision of rectum (APER).
Method All patients who underwent APER operations for rectal cancer in one NHS trust between Jan 2004 and Nov 2014 were identified from a pathology database. This covered the period in which MRI gradually replaced CT scanning in staging the pelvis. Exclusion criteria were benign disease, cancers other than adenocarcinoma and absence of CT or MRI scan preoperatively or where lymph node status was not reported. Pathological TNM staging was used to assess regional lymph node (LN) involvement. These findings were compared with those reported on CT and/or MRI scan.
Results 150 patients fulfilled the criteria. 132 patients had pre-operative CT scan assessment of lymph node status. 58 patients had pre-operative MRI scan assessment. 48 patients had both a CT and an MRI scan pre-operatively. Table 1shows the comparison of reported lymph node stage by histopathological assessment of the specimen and CT/MRI.
In terms of identifying LN positivity, CT achieved a sensitivity of 39% with a specificity of 78% while MRI achieved a sensitivity of 71% with a specificity of 35%.
Conclusion The specificity of MRI in this cohort of patients undergoing APER suggests that decisions based on LN positivity will risk significant overtreatment. Although CT was considerably less sensitive it was more specific. As MRI was gradually replacing CT, data were available for comparison but not from identical time periods. There were also a number of radiologists involved in reporting both the CT and MRI scans. However, these data do suggest that in routine practice, pre-operative imaging of low rectal cancer by CT will under-estimate LN involvement while reliance on MRI will over-estimate LN involvement.
Disclosure of interest None Declared.