Management of hepatocellular carcinoma (HCC) is frequently limited by liver cirrhosis and aggressiveness of the tumor. Despite the implementation of surveillance programmes and administration of antiviral therapy for cirrhotic population, more than 60 percent of HCC are not amenable to curative surgery or local ablation at presentation.[1-3] Palliative treatment for unresectable HCC is continuously evolving: cytotoxic chemotherapy is associated with a radiological response rate of 10 percent and a serological response of 30 percent[4, 5] but the treatment is not widely considered standard because of concern about toxicity in cirrhotic patients. Following two positive phase III trial results comparing transarterial chemoembolization (TACE) with supportive care,[6, 7] TACE is frequently recommended in patients without portal vein invasion and extra-hepatic disease. The recent development of the multi-targeted agent, sorafenib, has not only led to the approval of the agent for the treatment of unresectable HCC but has also resulted in an increasing list of novel targeted agents being tested for HCC in the clinical setting.[8, 9]
Selecting suitable subjects for drug testing in HCC is more challenging than in other solid tumors. Apart from tumor factors, clinicians need to consider non-tumor factors such as performance status and hepatic reserve to help in the prognostication of the patients. As a result, conventional TNM staging system, which takes into consideration only the tumorous factors, is inadequate for gauging the prognosis of HCC patients. Over the past decade a number of staging systems with various properties have been developed in different regions, with notable examples including the Cancer of the Liver Italian Program (CLIP), Barcelona Clinic Liver Cancer (BCLC), Chinese University Prognostic Index (CUPI) and the Advanced Liver Cancer Prognostic System (ALCPS) staging systems. CLIP was initially proposed by the Italian group for prognostication of HCC. Subsequent validation work has confirmed that CLIP is applicable in both Western and Asian HCC populations.[11, 12] BCLC is derived from Western HCC patients, and has been validated in different populations of HCC. The system is enhanced by the linkage of treatment algorithms to each tumor stage. CUPI and ALCPS have been reported by two Hong Kong groups for the prognostication of hepatitis B virus-related HCC. CUPI is constructed based on the combination of TNM system together with a number of statistically weighted parameters, with an aim of staging both operable and inoperable HCC. ALCPS is developed from patients with advanced HCC and is mainly used to predict the short-term outcome of patients.
It is clear that the population of unresectable HCC is a highly heterogeneous group of patients with a wide spectrum of survival, ranging from a few months only to more than 2 years.[16-18] To facilitate drug testing in clinical trials in advanced disease, an optimal staging system should be able to accurately predict the short-term survival of patients so that those with terminal disease are not subjected to complicated trial procedures and toxicity when there is little or no chance of benefit. Each staging system has its own definition of terminal disease. The BCLC system defines patients with Child's C cirrhosis or and an Eastern Cooperative Oncology Group performance status of 3 or higher as Stage D. According to the treatment algorithm, this group of patients has terminal disease and should be put on supportive care. For CUPI, the high-risk group has extremely poor prognosis with a 3-month survival rate of lower than 25 percent. As a result of such short overall survival, the CUPI high-risk group is considered not suitable for recruitment into clinical trials. For CLIP, most reports suggest that a CLIP score of 5 to 6 probably represents a homogeneous group of patients with terminal disease. For ALCPS, the poor prognostic group (score 16 or above) has a 3-month survival rate of 30 percent of lower, and this group of patients is again probably not suitable for undergoing drug testing in clinical trials.
In the current issue, Li et al. evaluate the ability of 12 staging systems for HCC to predict the short-term outcome of patients who are not amenable to surgery or locoregional therapy. By retrospectively reviewing a cohort of 208 patients and analyzing the survival data according to each staging system, Li et al. found that CLIP, CUPI and ALCPS significantly outperform other staging systems in predicting the 3-month survival rate. In addition, Li et al. report that the use of eligibility criteria in Sorafenib Hepatocellular Carcinoma Assessment Randomized Protocol (SHARP), the phase III clinical trial comparing sorafenib with placebo in a Western population, is suboptimal for predicting a 3-month survival rate. In particular, of the 55 patients who fulfilled the inclusion criteria of the SHARP study, 38 died within 3 months. On the other hand, of the 153 patients who met the exclusion criteria of the SHARP study, 56 survived longer than 3 months, resulting in an overall misclassification rate of 45.2 percent (38+56/55+153). The study by Li et al. carries two important messages. First, the results indicate that the empirical eligibility criteria adopted by modern clinical trials in HCC are largely unable to discriminate between suitable versus unsuitable patients for testing novel agents. A more precise system is required to identify a homogenous group to enable the prediction of overall survival and the calculation of sample size for clinical trials. Second, although the BCLC stage is recommended by Western guidelines as a standard staging system for HCC, the system is not sensitive enough to predict the 3-month survival rate, and it remains doubtful whether BCLC Stage C patients are the best group to be tested with novel systemic agents in Asia.
Should we generalize the data of Li et al. to routinely apply CLIP, CUPI or ALCPS in selecting patients for drug testing? There are several issues we need to consider: First, the study by Li et al. is limited by a relatively small sample size (N = 208). Also, the 3-month survival rate of the current study population of 45 percent is remarkably lower than the survival figure in SHARP study and other prospective HCC cohorts in Asia. It is highly likely that Li et al. have recruited patients with heavy prior treatment or at later course of disease. A prospective study with a larger sample size and less selective population is necessary to validate the findings of current study. In addition, Li et al. have dichotomized the study population in each staging system to simplify the comparative analysis. However, in real-life practice, each system stratifies patients into at least three risk groups. The study by Li et al. does not clearly indicate the most suitable categories of each staging system for recruitment into clinical trials. Therefore, the focus should on the evaluation of the performance of the CUPI intermediate stage, CLIP score 3–4 and the ALCPS intermediate group, respectively, in seeking to identify suitable subjects for clinical trial, before advocating the routine use of these staging systems.
Conventionally, owing to the lack of choice and the low efficacy of systemic agents, patients with unresectable intra-hepatic HCC are recommended TACE whenever possible before systemic agents are considered. At present, there has not been universal agreement on the maximal number of TACE cycles and the appropriate criteria to declare a TACE failure when changing to systemic agents may be beneficial to patients. Some clinicians prefer to continue TACE regardless of intra-hepatic progression or defer the decision to stop TACE until the emergence of contraindications to TACE. As a result, patients may have been treated with more cycles of TACE than justifiable and suffered disease progression to a state beyond which systemic agents could be considered. Also, according to phase III clinical trial data and various series, it is evident that patients with an extensive HCC burden (e.g. a tumor diameter >10 cm) derive significantly less benefits from transarterial therapy.[6, 21-23] Given the anticipated increasing number and potentially improved efficacy of systemic agents for HCC in future, studies are indicated to refine the eligibility of subjects for TACE and to define the best timing for shifting from TACE to systemic agents.
Finally, it is not infrequent to come across novel targeted agents demonstrating promising efficacy in preclinical models of HCC but failing to produce significant clinical benefits. This is most likely due to our incomplete understanding of the complexity of the mechanism and the key driver event involved in the progression of HCC. Experience from the lung and breast cancer fields has shown that success in clinical trials on targeted therapy can be improved only if we are able to apply an agent to a group of appropriately selected patients enriched by biomarkers. Although the driver mutations of HCC are not yet clearly understood, sequencing studies in both tissue and plasma samples have already deciphered some potential druggable molecular defects of HCC.[25, 26] In fact, recent phase II data on novel agents including c-Met inhibitors and histone deacetylase inhibitors show that biomarkers could potentially identify responding patients for HCC.[27, 28] Conventionally, biopsy tissue of HCC is usually not required or even not recommended in the design of clinical trials. In the current era of biomarker-driven development, it is imperative to obtain histological samples for biomarker analysis alongside evaluating the efficacy of novel agents. Preferably, the clinical evaluation of biomarkers should be available before proceeding to large-scale phase II or phase III clinical trials.
In conclusion, due to the complexity of disease, conventional drug development for HCC has been limited by a number of factors. The results of study by Li et al. have enlightened us to the importance of recruiting the appropriate patients for clinical trials testing novel drugs for HCC. The use of a proper staging system is helpful in the identification of patients with terminal disease who could then be spared unnecessary trial procedures. Apart from patient selection, it is crucial to look into the timing of switching from TACE to systemic agents and the need for a biopsy to explore tissue biomarkers in clinical trials. The development of a targeted agent for HCC will require the advance of knowledge in these three respects.