Reproducibility of mRECIST in assessing response to transarterial radioembolization therapy in hepatocellular carcinoma

The purpose of our study was to evaluate the reproducibility of Modified Response Evaluation Criteria in Solid Tumors (mRECIST) in hepatocellular carcinoma (HCC) lesions undergoing transarterial radioembolization (TARE) therapy and to determine whether mRECIST reproducibility is affected by the enhancement pattern of HCC. One hundred and three HCC lesions from 103 patients treated with TARE were evaluated. The single longest diameter of viable tumor tissue was measured by two radiologists at baseline; response to therapy was evaluated according to mRECIST. The enhancement pattern of HCC lesions was correlated with their mRECIST response. The response rate between mRECIST and RECIST 1.1 was compared. Wilcoxon signed-rank test, paired t test, Lin's concordance correlation coefficient (ρ_c), Bland-Altman plot, kappa statistics, and Fisher's exact test were used to assess intra- and interobserver reproducibilities and to compare response rates. There were better intra- than interobserver agreements in the measurement of single longest diameter of viable tumor tissue (bias=0 cm intraobserver versus bias=0.3 cm interobserver). For mRECIST, good intraobserver (K{green}=0.70) and moderate interobserver (K{green}=0.56) agreements were noted. The mRECIST response for HCC lesions with homogeneous enhancement at both baseline and follow-up imaging showed better intra- and interobserver agreements (K{green}=0.77 and 0.60, respectively) than lesions with heterogeneous enhancement at both scans (K{green}=0.54 and 0.40, respectively). In the early follow-up period mRECIST showed a significantly higher response rate than RECIST (40.8% versus 3.9%; P=0.025). Conclusions: In HCC patients treated with TARE, mRECIST captures a significantly higher response rate compared with RECIST; it also demonstrates acceptable intra- and interobserver reproducibilities for HCC lesions treated with TARE, and mRECIST reproducibility may be lower for HCC lesions with heterogeneous distribution of the viable tumor tissue.