Is dual-phase C-arm CBCT sufficiently accurate for the diagnosis of colorectal cancer liver metastasis during liver intra-arterial treatment?

Purpose: This study aimed to estimate the accuracy of dual-phase C-arm cone beam computed tomography (CBCT) for the detection of colorectal cancer liver metastases, as compared with multidetector computed tomography (MDCT).

Materials and methods: Between March 2014 and December 2016, 49 consecutive patients referred for intra-arterial treatment for colorectal cancer liver metastases were enrolled in a single-center observational study. All patients were examined with MDCT and with dual-phase C-arm cone beam computed tomography performed after iodine injection in the proper hepatic artery before intra-arterial treatment. Two blinded observers independently reviewed all examinations. Diagnostic accuracy was determined using both a six-cell matrix method and a "worst-case scenario."

Results: Readers identified at MDCT 264 colorectal liver metastases and 43 other liver lesions. The early and late arterial phase showed 240 and 277 liver lesions respectively. A certainty of the diagnosis was obtained in 63% and 85% at the early (EAP) and late arterial phase (LAP), respectively. Streak artifacts or liver segment truncation, or inadequate enhancement was responsible for the inability to see or to correctly adjudicate a lesion to a diagnosis in 27% and 15% of the cases at the EAP and LAP. The "worst-case scenario" yielded a Se and Sp of 58% and 51%, respectively, at EAP and 84% and 70%, respectively, at LAP.

Conclusion: On CBCT, EAP showed limited accuracy. LAP provided the best tumor detectability.

Key points: • The early arterial phase (EAP) yielded poor accuracy: Se = 58% and Sp = 51% (p < 0.0001). • The late arterial phase (LAP) phase yielded good accuracy: Se = 84% and Se = 70% (p = 0.02). • The probability of a correct diagnosis at the EAP was 60%.