BACKGROUND: The clinical significance of indeterminate pulmonary nodules (IPN) at staging computed tomography (CT) for colorectal cancer (CRC), and the optimal diagnostic approach, are debated. This study aimed to analyse variability in radiologists' detection of IPN at staging CT for CRC.

METHODS: All patients with CRC referred to our center between 2006 and 2011 were included. Primary staging CT scans were re-evaluated by an experienced thoracic radiologist whose findings were entered into a dedicated database and merged with data from the Danish Colorectal Cancer Group database, the National Patient Registry, the Danish Pathology Registry, and the primary CT evaluation. Inter-reader agreement was calculated by Kappa statistics, and associations between variables and malignancy of pulmonary nodules were analyzed with χ (2) and Mann-Whitney-Wilcoxon tests. Multivariable logistic regression analyses were used to adjust for potential confounding variables.

RESULTS: In total, 841 patients were included. The primary CT assessment reported IPN in 9.8 % of patients and pulmonary metastases in 5.1 % of patients compared with 5.6 and 7.0 %, respectively, reported by the experienced thoracic radiologist. Kappa for agreement between the primary assessor and the thoracic radiologist on IPN was 0.31 and 0.65 for pulmonary metastases. Synchronous liver metastases were predictive of malignancy of IPN (adjusted odds ratio 20.1; 95 % confidence interval 2.64-437.66; p = 0.012), whereas no other investigated radiological characteristics or clinicopathological factors were significantly associated with malignancy of IPN.

CONCLUSION: The characterization of pulmonary findings on staging CT for CRC varied greatly between the radiologists, and double-reading of scans with IPN is recommended prior to further diagnostic work-up.