Purpose: To assess the diagnostic accuracy of image subtraction compared with nonsubtracted images obtained with contrast-enhanced T1-weighted imaging (CE T1WI) for the diagnosis of hepatocellular carcinoma (HCC) necrosis after transarterial chemoembolization (TACE), using liver explant as the reference standard.
Materials and methods: Thirty-four patients who underwent TACE within 90 days of liver transplantation and CE MRI scans were assessed by two independent observers who determined the percentage of tumor necrosis using nonsubtracted and subtracted postcontrast phases. Histopathologic percentage of necrosis was retrospectively determined by an experienced pathologist. Spearman rank correlation test was used to correlate the percentages of necrosis from MR evaluation and from pathology. Receiver operating characteristics curve analysis was performed to determine the performance of subtracted versus nonsubtracted datasets for the diagnosis of complete tumor necrosis.
Results: There were 57 HCCs (mean size, 2.4 cm; range, 1.2-4.2 cm) diagnosed at explant and identified on MRI, including 16 completely necrotic HCCs. Subtraction demonstrated better interobserver agreement than nonsubtraction dataset for the diagnosis of tumor necrosis. There was a strong correlation between image subtraction and histopathologic assessment of necrosis (r = 0.80-0.86, depending on the phase, P < 0.0001). Subtraction demonstrated significantly higher sensitivity and accuracy for the diagnosis of complete tumor necrosis compared with nonsubtracted dataset.
Conclusion: Image subtraction enables accurate assessment of necrosis of HCC after TACE with the best accuracy observed at the arterial phase.