Abstract
In rectal cancer, preoperative staging should identify early tumours suitable for treatment by surgery alone and locally advanced tumours that require therapy to induce tumour regression from the potential resection margin. Currently, local staging can be performed by digital rectal examination (DRE), endoluminal ultrasound (EUS) or magnetic resonance imaging (MRI). Each staging method was compared for clinical benefit and cost-effectiveness. The accuracy of high-resolution MRI, DRE and EUS in identifying favourable, unfavourable and locally advanced rectal carcinomas in 98 patients undergoing total mesorectal excision was compared prospectively against the resection specimen pathological as the gold standard. Agreement between each staging modality with pathology assessment of tumour favourability was calculated with the chance-corrected agreement given as the kappa statistic, based on marginal homogenised data. Differences in effectiveness of the staging modalities were compared with differences in costs of the staging modalities to generate cost effectiveness ratios. Agreement between staging and histologic assessment of tumour favourability was 94% for MRI (kappa=0.81, s.e.=0.05; kappa(W)=0.83), compared with very poor agreements of 65% for DRE (kappa=0.08, s.e.=0.068, kappa(W)=0.16) and 69% for EUS (kappa=0.17, s.e.=0.065, kappa(W)=0.17). The resource benefits resulting from the use of MRI rather than DRE was 67164 UK pounds and 92244 UK pounds when MRI was used rather than EUS. Magnetic resonance imaging dominated both DRE and EUS on cost and clinical effectiveness by selecting appropriate patients for neoadjuvant therapy and justifies its use for local staging of rectal cancer patients.