Accuracy of pre-treatment locoregional rectal cancer staging in a national improvement project

Background: The aim of this study was to assess the accuracy, particularly the predictive value, of locoregional clinical rectal cancer staging (cTN) and its variability in a national improvement project. Methods: cTN stages and the distance between tumour and mesorectal fascia (MRF) were compared with histopathological findings in 1168 patients who underwent radical resection without neoadjuvant treatment. Data were registered prospectively from 2006 to 2014. Results: Agreement between clinical and histopathological TN stages was 50%, independent of tumour location. Inter-hospital variability was within 99% prediction limits. Magnetic resonance imaging (MRI) was increasingly applied, but staging accuracy did not improve. Stage II-III was correctly predicted in 69% and pStage I was over-staged in 35%. The positive predictive value of endorectal ultrasonography (ERUS) for T1 lesions was 57%. MRI-based distances to MRF correlated poorly with the circumferential resection margin (r = 0.26). A negative resection margin was achieved in 91% when the distance to the MRF was > 1 mm. Conclusions: The accuracy of rectal cancer staging in general practice should be improved to avoid under-or overtreatment. Training and expert review of pre-treatment MR imaging could be helpful. A second ERUS is justified when transanal local resection for early lesions is planned.