To the Editor:
We read with interest the paper comparing the 7 currently used staging systems for hepatocellular carcinoma (HCC) published in a recent issue of HEPATOLOGY.1 The authors concluded that the BCLC system, proposed by the Barcelona study group, has the best predictive power compared to other systems. The main reason why authors have reached this conclusion could be related to the distinct characteristics of the study population as a substantially high proportion of patients (21%) had undergone liver transplantation. This unique feature made BCLC a prevailing system because most transplant candidates complied with the Milan criteria (1 nodule ≤5 cm in diameter, or ≤ 3 nodules and ≤ 3 cm in diameter) which is also adopted in the BCLC system. In addition, another 32% of patients had received aggressive antitumor treatment (4% resection, 19% radiofrequency ablation and 9% chemoembolization). This feature again made BCLC a preferred model when the non-transplant patients were separately analyzed because BCLC contained these treatment-derived parameters. The authors' findings are generally consistent with other two independent studies which also confirmed the prognostic power of BCLC.2, 3 However, since the CLIP system has been prospectively validated and proposed as the primary staging system for HCC,4 it would be interesting to examine why entirely different conclusions were reached in different series.
The design and function of the staging system for HCC do not necessarily link with its treatment. The CLIP system was originally constructed based on unselected patient population in which the majority of patients belonged to an intermediate or late stage.5 With this feature, any active forms of treatment are less likely to perform, and certain important prognostic predictors, such as number and size of tumor which are major determinants associated with the outcome, appear to be less informative in terms of outcome prediction. Consistent with the current and previous two studies showing that BCLC is a more favorable model,1–3 the Japan Integrated System (JIS), which combines Child-Turcotte-Pugh score and tumor-node-metastasis system, is better than the CLIP system for outcome prediction.6 A major cause why JIS can beat the CLIP system is that the vast majority (96%) of patients had undergone radical treatment (resection or loco-regional therapies) in that study, suggesting most of them belonged to an early or intermediate stage. Therefore, the BCLC and JIS system are intrinsically similar because both models contain treatment-derived variables that may work well in an appropriate study environment when compared to the CLIP or other systems.
A potential shortcoming of the BCLC system is that the prognostic predictors are not properly weighted according to their influence. In contrast, the CLIP system, which only requires simple calculation at bedside assessment, has a wide score range from 0 to 6 (7 categories) that is expected to more likely differentiate the disease severity. It should be noted that the studies directed to compare the predictive power had different patient demographics, including the etiologies of underlying chronic liver diseases, and variable treatment strategy that could subsequently impact the their predictive ability.7 Therefore, any staging system that proves to work well in a certain patient population needs to be tested in other populations with entirely different characteristics from early to advanced stage before it can be justified as the best tool for prognostic prediction.