An investigation on gastric cancer staging using CT structured report

Objective: To evaluate the clinical performance of structured report (SR) for CT in patients with pre-operative staging of gastric cancer, compared to non-SR.

Methods: Retrospectively, 51 consecutive cases with primary gastric cancer staging were enrolled. Every SR or non-SR was performed by two GI radiologists (1 junior and 1 senior). Interobserver agreement was conducted between the junior and senior groups for both SR and non-SR. 10 key features required for lesion description and staging were assessed between SR and non-SR. Diagnostic content between SR and non-SR was also compared. Accuracy of SR and non-SR on T staging was measured. Subjective evaluation of SR vs. non-SR was also conducted in form of survey by 20 radiologists and 3 GI surgeons.

Results: Interobserver agreement showed excellent in SR (Kappa = 1, P < 0.001), but poor in non-SR (Kappa = 0.036, P = 0.455). For the 10 key features required for lesion assessment, non-SR showed 6.84 ± 0.83 while SR reported all of them (P < 0.001). Statistically significant improvement was observed in the SR for parts of key features, especially for assessment of adjacent organs and vessels (P < 0.001). Accuracy comparison of T staging showed higher in SR for cohort of T4a (P = 0.028<0.05). The scores of subjective evaluation were higher (P < 0.05) in SR than in non-SR by both radiologists and surgeons. Meanwhile, the inter-observer agreement among surgeons was good in SR with significance (w=0.53, P = 0.005 for efficiency; w=0.638, P < 0.001 for integrity) but poor in non-SR.

Conclusions: SR of gastric multiphasic CT ensured reliable detection of all relevant key features for staging along with reproducible documentation, which was not always the case for non-SR. In addition, SR has the potential in improving diagnostical accuracy of T staging and was welcomed by clinicians.