Absence of Tumor Invasion into Pelvic Structures in Locally Recurrent Rectal Cancer: Prediction with Preoperative MR Imaging

Abstract

To retrospectively assess the accuracy of preoperative magnetic resonance (MR) imaging for identification of tumor invasion into pelvic structures in patients with locally recurrent rectal cancer scheduled to undergo curative resection. The institutional review board approved this study, and informed consent was waived because of the retrospective nature of the study. Preoperative MR images in 40 consecutive patients with locally recurrent rectal cancer scheduled to undergo curative treatment between October 2003 and November 2006 were analyzed retrospectively. Four observers with different levels of experience in reading pelvic MR images assessed tumor invasion into the following structures: bladder, uterus or seminal vesicles, vagina or prostate, left and right pelvic walls, and sacrum. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated, and a receiver operating characteristic curve was constructed. Surgical and/or histopathologic findings were used as the reference standard. Interobserver agreement was measured by using kappa statistics. Preoperative MR imaging was accurate for the prediction of tumor invasion into structures with negative predictive values of 93%-100% and areas under receiver operating characteristic curves of 0.79-1.00 for all structures and observers. Positive predictive values were 53%-100%. Disease was overstaged in 11 (observer 1), 22 (observer 2), 10 (observer 3), and nine (observer 4) structures and was understaged in nine (observer 3) and two (observer 4) structures. Assessment failures were mainly because of misinterpretation of diffuse fibrosis, especially at the pelvic side walls. Interobserver agreement ranged between 0.64 and 0.99 for experienced observers. Preoperative MR imaging is accurate for the prediction of absence of tumor invasion into pelvic structures. MR imaging may be useful as a preoperative road map for surgical procedure and may thus increase chances of complete resection. Interpretation of diffuse fibrosis remains difficult.