Aim A pilot study was undertaken to determine the accuracy of computed tomography (CT) staging in identifying patients with high-risk colon cancers who would be considered as candidates for a neoadjuvant therapy trial (FOxTROT) and those at low risk (T1/T2) who would be excluded.
Method Participating radiologists from 19 centres attended workshops for standardization of image interpretation according to previously defined prognostic criteria: good prognosis tumours, including, T1/T2; intermediate prognosis, T3 < 5 mm tumour invasion beyond the muscularis propria (MP); and poor prognosis tumours, including T3 with tumour extension ≥ 5 mm beyond the MP or T4. The CT findings were compared with histopathology as the reference standard.
Results Of 94 patients with radiological and pathological data, 71% were categorized by CT as having a poor prognosis. The sensitivity and specificity of CT in identifying these tumours were 87% (95% CI, 74–94) and 49% (95% CI, 33–65). Sensitivity and specificity for tumour infiltration beyond the MP (T3/T4 vs T1/T2) were 95% (95% CI, 87–98) and 50% (95% CI, 22–77), respectively. Including all CT-staged T3 and T4 patients in the trial would have increased the proportion eligible for entry to 89% (n = 84) without affecting the false-positive rate of 7%. Some 20% of T3/T4 patients would have been ineligible for FOxTROT because of synchronous metastases.
Conclusion In a multicentre setting, CT scanning identified high-risk (T3/4) colon cancers with minimal overstaging of T1/T2 tumours, thus establishing the feasibility of radiologically guided neoadjuvant chemotherapy.