CT staging performance in an international trial of neoadjuvant chemotherapy for locally advanced colon cancer

Objectives

In FOxTROT, neoadjuvant chemotherapy (NAC) significantly reduced recurrence risk, compared to upfront surgery, in locally advanced colon cancer. This analysis evaluates the correlation between radiological and pathological staging within the trial to support the adoption of CT-based patient selection.

Methods

In this pre-planned analysis of prospectively collected data, local radiological and pathological staging were compared in upfront surgery participants. T stage, N stage and extramural venous invasion (EMVI) status were evaluated using overall agreement, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Subgroup analyses explored the impact of mismatch repair status and tumour side.

Results

354 participants were included. T stage agreement was 63.0%; T3 and T4 tumours were correctly identified in 78.9% and 41.1% of participants, respectively. The PPV for T3-4 status was 94.5%. N stage agreement was 39.8%; for N status (positive vs. negative), overall agreement, sensitivity, specificity, PPV and NPV were 54.1%, 81.1%, 26.0%, 53.2% and 57.1%, respectively. For EMVI, these values were 54.9%, 71.0%, 41.2%, 50.7%, and 62.5%, respectively. Accuracy metrics did not differ significantly by tumour side or mismatch repair status.

Conclusions

CT effectively predicted T3-4 status with minimal over-staging, but performed poorly for individual T stage, N stage and EMVI. We propose radiological T3-4 status should be adopted as the primary biomarker for neoadjuvant patient selection, with molecular biomarkers to guide treatment choice.

Advances in Knowledge

In this multi-centre trial, local radiologists accurately identified T3-4 status to select participants for NAC, indicating utility for future neoadjuvant trials and clinical practice.