Abstract
Abstract only
e15653
Background: Several staging systems and models (TNM, BCLC, Okuda, CLIP and ALBI) have been developed to estimate the prognosis of patients with HCC. Most of these were developed prior to the prevalent use of sorafenib. The purpose of this study was to compare the prognostic and discriminatory power of these models in predicting survival for HCC patients treated with sorafenib. Methods: Patients who received sorafenib for the treatment of HCC between January 1, 2008 and June 30, 2015 in the provinces of British Columbia and Alberta, as well as Princess Margaret Cancer Centre and Sunnybrook Odette Cancer Centre in Toronto, Ontario were included. Survival outcomes for each model were assessed with Kaplan-Meier (KM) curves and compared with the log-rank test. Time dependent area under the curve (t-AUC) was used to test the discriminatory power of each model (higher t-AUC = more discriminatory power). Akaike information criterion (AIC), a measure of goodness-of fit of models while penalizing overly complex models, was used to compare the models (lower AIC = better model). Results: A total of 681 patients were included in this analysis. Median age was 64 years (range 8-91). Majority were males (80%), had a Child-Pugh score A (86%), ECOG performance status 0 (30%) and 1 (60%). 37% of patients were of East Asian ethnicity. Most common etiology for liver disease was hepatitis B (33%) and C (29%). At start of sorafenib, most patients were BCLC stage C (92%) and TNM stage IV (61%). The median overall survival for the entire cohort was 9.2 months (95% CI 8-10.4). See table below for t-AUC and AIC results. Conclusions: According to ourlarge multi-center study, CLIP appears to be the most informative in predicting survival in HCC patients treated with sorafenib. Prospective studies are needed to determine its role in patient selection for clinical trials and in guiding treatment decisions. The TNM and BCLC staging systems were the least useful in predicting survival in this population. [Table: see text]