ABSTRACT
AIM: To compare contrast-enhanced subtraction magnetic resonance imaging (MRI) with contrast-enhanced standard MRI in assessing treatment response following loco-regional therapies for hepatocellular carcinoma (HCC). METHOD AND MATERIALS: Institutional review board approval was obtained and informed consent was waived for this retrospective study. All patients were analysed from our institution's liver tumour database that had loco-regional HCC therapy and the following: (1) a contrast-enhanced MRI ≤6 weeks post-treatment, (2) an unenhanced T1-weighted high-signal treatment zone (TZ) ≥1 cm, (3) follow-up contrast-enhanced MRI performed ≥6 months post-treatment. Randomized standard and subtraction TZ datasets were independently assessed by three blinded radiology readers for either complete treatment necrosis or residual disease. The standard of reference (SOR) comprised a consensus read by two radiologists with knowledge of the follow-up MRI and all available clinical data. Statistical analyses were performed using receiver operating characteristics (ROC), t-test, and kappa statistic. RESULTS: Twenty-six patients (19 male and seven female patients; mean age 60 years, standard deviation 10.9 years, range 46-88 years) had a total of 45 corresponding HCCs and TZs. For ROC, the area under the curve (AUC) was 0.93 (subtraction protocol) versus 0.90 (standard protocol; p = 0.49). For the t-test, the mean reader confidence level was 4.4, 3.6, and 4.4 (subtraction protocol) versus 3, 3, and 3.7 (standard protocol; p ≤ 0.011). The kappa statistic for reader-to-SOR agreement was 0.83, 0.63, and 0.71 (subtraction protocol) versus 0.51, 0.36, and 0.64 (standard protocol). CONCLUSION: Subtraction MRI significantly improves the reader confidence level in the assessment of treatment response following loco-regional therapies for HCC.
