Many staging systems for hepatocellular carcinoma: Evolution from Child‐Pugh, Okuda to SLiDe

See article in J. Gastroenterol. Hepatol. 2004; 19: 805–811. Clinical staging systems for cancer patients are designed to give a more accurate prognostic assessment and to guide decisions in planning optimal treatment strategies. Current advances in medical examinations and medical care for patients with hepatocellular carcinoma (HCC) have resulted in a dramatically increased patient survival rate.1–3 However, despite such progress, making a reliable estimate of the prognosis of patients with HCC still remains difficult. It is evident that the prognostic staging system is an important clinical issue in deciding whether to treat a patient aggressively, avoid over-treatment or choose the palliative care option for patients with HCC. Therefore, recently several models designed to more precisely predict the outcome of patients with HCC have been proposed. However, the problems regarding the appropriate weights given to the variables measuring the residual liver function according to their relative importance, and how tumor-related characteristics are statistically derived and then applied to a prognostic model for HCC, to more easily and more accurately predict the outcome, still remain. The ideal systems to estimate the prognosis of patients diagnosed with HCC should be both simple and have a high discriminatory ability. There have been a large number of predictive factors that are expected to contribute to the prognosis of HCC. It is generally agreed that four main predictors exist regarding survival in HCC: (i) tumor stage; (ii) residual liver function; (iii) general health of patients; and (iv) the specific intervention.4 The classifications which are most commonly used for HCC are the Child-Pugh5 score, the Okuda Classification6 and tumor node metastasis (TNM) staging7 (Table 1). However, each has its own problems and limitations. The Child-Pugh score only accounts for variables reflecting the liver functional characteristics, not for tumor characteristics, because it does not include cancer parameters. Moreover, the quantification of the encephalopathy and ascites cases evaluated by the Child-Pugh score are subjective and lack precision. Conversely, the TNM classification includes only features related to the tumor and does not include liver function parameters. The TNM classification, which has become widely used for hepatic resection or orthotopic liver transplantation since the article by Mazzaferro et al.,8 has been found to be inadequate by many groups over the last few years.9,10 The TNM classification was not significantly correlated with the survival rate in patients undergoing a single treatment of transcatheter arterial chemoembolization,8 resection11 or orthotopic liver transplantation.8,10 A potential criticism of the TNM classification is that it requires a tumor biopsy to determine the grade. In addition, because more than 90% of patients with HCC do not undergo surgery3 this classification can rarely be used. The Okuda Classification, which was the first prognostic model used for approximately two decades, includes parameters related to liver function status such as albumin, ascites and bilirubin, and to the tumor stage such as 50% of liver parenchymal area being involved. However, it does not include important tumor factors such as whether the tumor is unifocal, multifocal or diffuse, whether there is portal vein invasion, or how high the tumor markers are, all of which have prognostic significance. The Okuda Classification uses a high threshold for bilirubin of 3 mg/dL, which is probably unsuitable for the current population of HCC patients because many are now diagnosed early during the asymptomatic stage of disease due to the implementation of established screening programs in high-risk populations.1–3,12 The classic staging systems have been reevaluated, and more accurate staging systems for HCC have recently been reported from Italy, France, Spain, China and Japan. These include the Cancer of the Liver Italian Program (CLIP) scoring system,13 the French classification,14 Barcelona Clinic Liver Cancer (BCLC) staging,15 the Chinese University Prognostic Index (CUPI),16 and the Japan Integrated Staging (JIS) score system17 (Table 1). These newer classifications include prognostic factors which show statistical significance by multivariate analyses. The CLIP staging system combines Child-Pugh grade with tumor characteristics including tumor morphology, portal vein thrombosis and alpha-fetoprotein (AFP) levels. The French classification system consists of five prognostic factors: Karnofsky performance status index, serum bilirubin, serum alkaline phosphatase, serum AFP and ultrasonographic portal obstruction. The BCLC staging system includes the Child-Pugh classification, Okuda Classification, single or multifocal tumors, vascular invasion, portal hypertension, and performance status. The CUPI staging system is constructed by adding such factors as total bilirubin, ascites, alkaline phosphatase, AFP and asymptomatic disease into the TNM classification. Conversely, the JIS staging system combines Child-Pugh grade and TNM staging by the Liver Cancer Study Group of Japan (LCSGJ).18 However, these new staging systems still have some limitations. The CLIP staging system includes parameters related to the tumor stage, such as approximately 50% of the liver parenchymal area being involved, which is probably unsuitable for the current population of HCC patients because many are now diagnosed when the tumor size is still small because of the implementation of established screening programs in high-risk populations.1–3,12 Also, it cannot discriminate advanced stage cases such as those with a CLIP score from 4 to 5, and has a discrepancy between the tumor morphology and prognosis, and it classifies more than two-thirds of patients in the early stage (CLIP score 0–1), which is unsuitable for HCC patients diagnosed at an earlier stage. In the French classification, serum alkaline phosphatase is selected as an independent prognostic factor, which is less widely recognized, although serum alkaline phosphatase may be mainly related to the growth rate of HCC,19 and it includes subjective evaluation factors such as performance status, which is difficult to determine and analyze retrospectively. In the BCLC staging system, it is initially developed for patients undergoing a surgical resection for HCC, therefore the explication of the treatment-decision algorithm rather than a prognostic system; it includes subjective evaluation factors such as performance status and portal hypertension, which are not routinely measured factors to evaluate practically and are also questionably adequate factors in the prognosis in HCC. In the CUPI staging system limitations are that it includes a subjective evaluation factor such as 'asymptomatic disease on presentation'. It uses serum alkaline phosphatase as an independent factor and comprises highly advanced HCC patients whose 2- and 3-year survivals are 6.2% and 2.2%, respectively; and it comprises too many hepatitis B virus (HBV)-related HCC cases (79%) and uses the TNM classification for patients in whom the stage is difficult to determine. In the JIS staging system it needs to use the TNM staging by LCSGJ, including four tumor grades (stages I, II, III and IV) and liver damage grades (A, B and C), which have not been commonly used worldwide, and it cannot discriminate advanced stage cases such as those with a JIS score of 4–5. Regarding the prognostic factors selected in these new staging systems, regional variations in disparities between Western countries and Eastern countries creates controversy. Levy and Sherman20 stated that the CLIP staging system for HCC was easy to implement and more accurate than the Okuda Classification, as confirmed by a total of 257 Canadian patients with HCC. Similarly, Farinati et al. in Italy reported that the CLIP score can identify patients with different prognoses, particularly in the early phases of HCC, compared to the Okuda and TNM staging systems.21 Ueno et al. analyzed a total of 662 Japanese patients with HCC and concluded that the CLIP staging system had higher stratification ability with regard to prognosis in patients with HCC with every type of treatment than either the Okuda stage or the TNM classification.22 In contrast, Kudo et al. emphasized the advantage of their proposed new staging system, the JIS staging system, to have a high discriminatory value not only in early phase HCC patients but also in intermediate- and advanced-phase HCC patients.17 However, one of the criticisms of the CLIP, French, BCLC, CUPI and JIS staging systems is that they lack the measurement of des-gamma-carboxyl prothrombin (DCP) in the serum as a predictive factor. With respect to DCP, a new staging system for HCC proposed by Kawakita et al.23 in 2003 and a SLiDe (stage, liver damage and DCP) score termed by Omagari et al.24 in the July (2004) issue of the Journal are both worthy of special mention. Based on a multivariate model for predicting the outcome in the patients with HCC, DCP was chosen as one of the stronger prognostic factors in both new staging systems, while the serum AFP levels impacted on survival in the new staging system proposed by Kawakita et al. and did not impact on survival in the other. After classifying the patients according to the newer staging systems, which resulted in 141 patients and 177 patients with HCC, respectively, Kawakita et al.23 and Omagari et al.24 confirmed their new staging systems, including DCP, to be statistically better models for predicting the outcome in comparison to the CLIP or JIS scoring systems. However, both new staging systems also have several drawbacks. Kawakita et al. developed the new staging system based on an analysis of a small number of patients. In particular, this system comprised a small number of advanced HCC patients because, in all HCC patients analyzed, those with AFP levels greater than 400 ng/mL and those with portal thrombosis consisted of only 12.8% and 5.0% of all patients, respectively. In addition, they deleted portal vein thrombosis from the variables in their analysis. The other system using the SLiDe score24 has too many predicting factors (11), which are difficult to use in practice. Regarding simplicity, the score in Kawakita et al.'s staging system is easier to obtain than that in the SLiDe system. Recently, the kit used for DCP has been improved and now has increased sensitivity.25,26 Furthermore, recent reports have shown that high DCP serum levels are associated with larger-sized HCC27,28 and with portal thrombosis in patients with HCC.28,29 As described in these papers, DCP thus provides additional prognostic information which is expected to help in the development of new models for patients with HCC. Interestingly, both the CLIP staging system13 and the French classification14 include portal thrombosis as one of the independent predictive factors of survival, while DCP proved to be associated with portal thrombosis in HCC patients.28,29 Therefore, serum DCP as a predictive marker of survival in HCC might be potentially useful as a replacement for portal thrombosis. We need further examinations in the future to confirm these staging systems, including DCP, before any definitive conclusions can be made. A model to predict survival in patients with end-stage liver disease (MELD) was recently proposed to require a more refined scale that accurately represents the disease severity for liver transplantation.30,31 This system lacks tumor factors because it was developed for patients on the waiting list for transplant candidates. However, it should be noted that MELD consists of four variables, including serum bilirubin, creatinine levels, international normalized ratio (INR) for prothrombin time, and etiology of liver disease.30,31 As the serum creatinine level is known to frequently increase in end-stage liver disease, it has no potential as a predictor in the early stages of HCC. Schepke et al.32 recently reported that the MELD model is slightly superior to the Child-Pugh classification for the prediction of long-term survival in transjugular intrahepatic portosystemic shunt (TIPS) patients, and is better than the Child-Pugh classification in predicting short-term (3 months) outcome.33 However, whether or not the etiology of liver disease, which is not included in either the CLIP, French, BCLC, CUPI, JIS, a new staging system including DCP, or the SLiDe staging system, is a reliable measure of residual liver function in HCC patients remains to be discussed. Generally, regarding the limitations of the known staging systems, the process of developing a new staging system involves identifying potential factors depicting both liver function and tumor-related characteristics. The scoring system for HCC should consist of variables that are routinely assessed during a clinical examination. However, the new staging systems seem to often rely on additional parameters (Table 1) because a greater number of predictive parameters are analyzed by a multivariate model based on previously developed variables, which thereafter results in a moderate number of parameters being selected. Most of the new prognostic classifications are retrospective, recruit a reduced number of patients, or mix treated with untreated individuals diagnosed at disparate stages. Indeed, the CLIP staging system13 resulted in the following indices being incorporated; the tumor burden, such as the tumor morphology, portal vein thrombosis and the AFP level in addition to the Child-Pugh grade, in spite of the fact that a multivariate analysis is used. Therefore, new staging systems tend to be more complicated. Similarly, the new staging system developed by Omagari et al.24 is cumbersome to use in practice because of the 11 clinical parameters that need to be analyzed (tumor number, tumor size, tumor vascular involvement, lymph node metastasis, distant metastasis, ascites, serum bilirubin, serum albumin, indocyanine green retension test, prothrombin time, serum DCP), while the CLIP classification13 uses eight clinical parameters, the French classification14 uses five, the BCLC15 staging system uses nine, the CUPI uses six and the new staging system including DCP23 uses eight. As a result, many of the modifications of current systems may not become widely accepted because of their complexity. In addition, five of these systems (BCLC, CLIP, JIS, the new staging system including DCP and SLiDe score) use the Child-Pugh score whereas the French classification system and CUPI staging system do not. As a result, some of the Child-Pugh criteria may not be necessary. Other candidates for use as predictive variables of the survival of HCC have been suggested such as race; etiology of HCC; pathological grade of HCC; new laboratory or pathological parameters, especially those regarding metastasis, angiogenesis or the proliferation of tumors such as E-cadherin,34 vascular endothelial growth factor,35 hepatoma-derived growth factor,36 cyclin-dependent kinase,37 or the PCNA-labeling index.38 By analyzing the data, including these theoretical variables, from a large number of patients based on a multivariate analysis, we can possibly develop new models in the future that can be accurately applied to a large series of patients to increase the accuracy of the confidence limits. Another important problem is that numerous unsolved statistical questions remain regarding whether or not therapeutic modalities should be included in the survival model. The selection of patients according to the criteria of existing models has not been well defined. The periods during which the patients were analyzed by the various systems are as follows: CLIP score range was evaluated from 1990 to 1992; French classification from 1990 to 1992; CUPI from 1996 to 1998; the new staging system including DCP by Kawakita et al. from 1992 to 2001; and the SLiDe score from 1991 to 2002. Therefore, current specific treatments such as radiofrequency ablation or liver transplantation for HCC are likely to influence the outcome in either early- or intermediate-stage HCC patients, although treatments for advance-staged HCC may be ineffective even with aggressive treatment. As the selection of optimal therapeutic modalities could positively influence the outcome of patients, we need to develop a scoring system which stratifies the patients within therapeutic trials. Because therapeutic development for HCC in the future will possibly change patient prognoses, parameters that can change easily after treatment are not ideal. We thus need to establish a simple, precise, long-term tolerable and easy to use staging system to achieve a close correlation between the actual survival and predicted survival, thereby leading to the widespread use of such a system.39–41

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.