RESUMO
PURPOSE: Adequate staging of early rectal neoplasms is essential for organ-preserving treatments, but magnetic resonance imaging (MRI) frequently overestimates the stage of those lesions. We aimed to compare the ability of magnifying chromoendoscopy and MRI to select patients with early rectal neoplasms for local excision. METHODS: This retrospective study in a tertiary Western cancer center included consecutive patients evaluated by magnifying chromoendoscopy and MRI who underwent en bloc resection of nonpedunculated sessile polyps larger than 20 mm, laterally spreading tumors (LSTs) [Formula: see text] 20 mm, or depressed-type lesions of any size (Paris 0-IIc). Sensitivity, specificity, accuracy, and positive and negative predictive values of magnifying chromoendoscopy and MRI to determine which lesions were amenable to local excision (i.e., [Formula: see text] T1sm1) were calculated. RESULTS: Specificity of magnifying chromoendoscopy was 97.3% (95% CI 92.2-99.4), and accuracy was 92.7% (95% CI 86.7-96.6) for predicting invasion deeper than T1sm1 (not amenable to local excision). MRI had lower specificity (60.5%, 95% CI 43.4-76.0) and lower accuracy (58.3%, 95% CI 43.2-72.4). Magnifying chromoendoscopy incorrectly predicted invasion depth in 10.7% of the cases in which the MRI was correct, while magnifying chromoendoscopy provided a correct diagnosis in 90% of the cases in which the MRI was incorrect (p = 0.001). Overstaging occurred in 33.3% of the cases in which magnifying chromoendoscopy was incorrect and 75% of the cases in which MRI was incorrect. CONCLUSION: Magnifying chromoendoscopy is reliable for predicting invasion depth in early rectal neoplasms and selecting patients for local excision.