PRELIMINARY EXPERIENCE WITH THE MAGNETIC-RESONANCE ENDOSCOPE

Background: In contrast to endorectal surface coils used to assess pel vic tumors, the magnetic resonance endoscope (MR) has all the features of a standard endoscope. In ex-vivo imaging of the porcine gastrointe stinal tract, endoscopic MR demonstrates distinct histological layers of the gastrointestinal wall. The aim of this study was to assess the feasibility and the accuracy of endoscopic MR in local staging of pati ents with esophageal and rectal cancer. Methods: From April to August 1996, 12 patients (5 female and 7 male, mean age 63 [range 44-84] year s) with histologically proven esophageal (n = 6) and rectal (n = 6) ca ncer prospectively underwent endoscopic ultrasound (EUS) followed by e ndoscopic MR. The two radiologists reviewing the endoscopic MR images were blinded to the EUS results. Assessment of T and N stages was comp ared to EUS and histology. Results: Endoscopic MR was well tolerated i n all patients and there were no complications. Image quality was suff icient in 75%. Endoscopic MR T-staging correlated with EUS and histolo gy in 7/12 and 5/7 patients respectively. Discordance was due to overs taging by endoscopic MR. N-staging correlated with EUS and histology i n 10/12 and 6/8 cases respectively. Non-correlation was due to a numbe r of false negative results at endoscopic MR. Conclusion: These early results demonstrate endoscopic MR to be feasible and to produce compar able local staging to EUS in patients with esophageal and rectal cance r. The ultimate goal will be to combine endoscopic MR with body coil M R imaging (for the assessment of distant metastases) in order to provi de ''one-step staging'' for the entire evaluation of gastrointestinal tumors.