MRI surveillance for the detection of local recurrence in rectal cancer after transanal endoscopic microsurgery

Abstract

Objectives: To evaluate diagnostic performance of follow-up MRI for detection of local recurrence of rectal cancer after transanal endoscopic microsurgery (TEM). Methods: Between January 2006 and February 2014, 81 patients who underwent TEM were included. Two expert readers (R1 and R2), independently evaluated T2-weighted (T2W) MRI and diffusion-weighted (DWI) MRI for the detection of local recurrence, retrospectively, and recorded confidence on a five-point scale. Diagnostic performance of follow-up MRI was assessed using ROC-curve analysis and kappa statistics for the reproducibility between readers. Results: 293 MRIs were performed, 203 included DWI. 18 (22%) patients developed a local recurrence: luminal 11, nodal two and both five. Areas under the curve (AUCs) for local recurrence detection were 0.72 (R1) and 0.80 (R2) for T2W-MRI. For DWI, AUCs were 0.70 (R1) and 0.89 (R2). For nodal recurrence AUCs were 0.72 (R1) and 0.80 (R2) for T2W-MRI. Reproducibility was good for T2W-MRI (κ0.68 for luminal and κ0.71 for nodal recurrence) and moderate for DWI (κ0.57). AUCs and reproducibility for recurrence detection increased during follow-up. Conclusions: Follow-up with MRI after TEM for rectal cancer is feasible. Postoperative changes can be confusing at the first postoperative MRI, but during follow-up diagnostic performance and reproducibility increase. Key Points: • Follow-up with MRI is feasible for follow-up after TEM for rectal cancer. • DWI-MRI is a useful addition to detect recurrences after TEM. • Postoperative changes can be confusing and can lead to underestimation of recurrence. • Appearance of intermediate signal at T2W-MRI is suspicious for recurrence. • Nodal staging remains challenging.