ABSTRACT

In this study we aimed to evaluate the accuracy of preoperative rectal cancer staging in our center, using 1.0 Tesla Magnetic resonance imaging (MRI) with phased-array coils and determine if the results can be improved by adopting thin-section MRI techniques.

Eighty-four patients with biopsy proven rectal cancer were prospectively evaluated by MRI using either the standard (8 mm sections in all planes) or the thin-section protocol (additional 5 mm sections in oblique axial plane perpendicular to the long axis of the tumor). Patients undergoing surgery with or without neoadjuvant therapy (standard MRI, n=15 and thin-section MRI, n=22) were included in the analysis. TNM stage, circumferential resection margin (CRM) and adjacent organ involvement were compa-red with histopathologic findings.

In the thin-section group, estimation of tumor stage was 59% accurate, showing fair agreement with histopathology (κ=0.38, p<0.05); sensitivity and specificity rates were 100% and 59% for T2 and 47% and 100% for T3 stages. In the standard group estimation of T stage was only 40% accurate, showing no significant agreement with histopathology. The most frequent staging error was under-staging of borderline pT3 tumors in both groups. Accuracy, sensitivity and specificity of positive CRM were 67%, 67% and 95% for thin-section group and 50%, 100% and 93% for standard group, respectively; agreement with histopathology was significant only for the thin-section group (κ=0.61; p<0.05). Accuracy of nodal status was 77% in the thin-section group (κ=0.51; p<0.05) and 87% in the standard group (κ=0.75; p<0.05).

Thin-section MRI techniques can be used to improve tumor staging and positive CRM predictions even with low field magnet systems.

Keywords: Magnetic Resonance Imaging, Rectal Cancer, Tumor Staging