Abstract
Advanced hepatocellular carcinoma (HCC) management has become more complex as novel therapies have been proven effective. After sorafenib, the approval of other multikinase inhibitors (MKIs) and immune checkpoints inhibitors (ICIs) has considerably increased the number of systemic therapies available. Therefore, careful assessment and monitoring of response to systemic treatment are essential to identify surrogate endpoints of overall survival (OS) in clinical trials and reliable tools to gauge treatment benefit in clinical practice. Progression-free survival (PFS) and objective response rate (ORR) are early informative parameters of efficacy that are not influenced by further lines of therapy. However, none of them has shown sufficient surrogacy to be recommended in place of OS in phase 3 trials. With such a wealth of therapeutic options, the prime intent of tumor assessments is no longer limited to identifying progressive disease to spare ineffective treatments to non-responders. Indeed, the early detection of responders could also help tailor treatment sequencing. Tumor assessment relies on the Response Evaluation Criteria for Solid Tumors (RECIST), which are easy to interpret - being based on dimensional principles - but could misread the activity of targeted agents. The HCC-specific modified RECIST (mRECIST), considering both the MKI-induced biological modifications and some of the cirrhosis-induced liver changes, better capture tumor response. Yet, mRECIST could not be considered a standard in advanced HCC. Further prognosticators including progression patterns, baseline and on-treatment liver function deterioration, and baseline alpha-fetoprotein (AFP) levels and AFP response have been extensively evaluated for MKIs. However, limited information is available for patients receiving ICIs and regarding their predictive role. Finally, there is increasing interest in incorporating novel imaging techniques which go beyond sizes and novel serum biomarkers in the advanced HCC framework. Hopefully, multiparametric models grouping dimensional and functional radiological parameters with biochemical markers will most precisely reflect treatment response.