RESUMO
AIM: The preoperative stage of rectal cancer is an important prognostic factor affecting treatment recommendations. Currently, magnetic resonance imaging (MRI) is used for clinical staging to identify patients who should be recommended for neoadjuvant oncological treatment. The aim of this work was to investigate the accuracy of the preoperative T- and N-stage and also involvement of the mesorectal fascia (MRF) as determined by MRI in a Swedish national cohort. METHOD: Patients who had undergone resectional surgery for rectal cancer without neoadjuvant treatment in Sweden in the period 2013-2017 were identified through the Swedish Colorectal Cancer Registry (n = 2062). The T- and N-stage determined by preoperative MRI were compared with the histopathological results. The MRI-assessed MRF status was compared with the circumferential resection margin. RESULTS: Sensitivity for differentiating T1-2 from T3-4 was 69% and specificity 77%. Sensitivity to detect N1-2 was 42% and specificity 81%. Sensitivity for MRF positivity was 50% and specificity 92%. Agreement analysed by weighted kappa analysis was 0.47 for T-stage [confidence interval (CI) 0.44-0.51], 0.24 for N-stage (CI 0.19-0.24) and 0.20 for MRF status (CI 0.12-0.29). CONCLUSION: In this study, accuracy of the preoperative MRI was lower than expected for rectal cancers. Overstaging might lead to potentially harmful neoadjuvant treatment and understaging can lead to a higher risk of tumour recurrence. The results of this study show that efforts should be made to increase the accuracy of the preoperative evaluation in order to optimize treatment recommendations.