- © 2003 by American Society of Clinical Oncology
Estrogen Receptor Classification for Hepatocellular Carcinoma: Comparison With Clinical Staging Systems
- Erica Villa,
- Alessandra Colantoni,
- Calogero Cammà,
- Antonella Grottola,
- Paola Buttafoco,
- Roberta Gelmini,
- Ilva Ferretti and
- Federico Manenti
- From the Division of Gastroenterology and Department of Surgery, University of Modena and Reggio Emilia, Modena; and Istituto Metodologie Diagnostiche Avanzate, Consiglio Nazionale delle Ricerche, and Gastroenterology Unit, University of Palermo, Palermo, Italy.
- Address reprint requests to Erica Villa, MD, Department of Internal Medicine, Division of Gastroenterology, University of Modena and Reggio Emilia, Via del Pozzo 71, 41100 Modena, Italy; email: villa.erica{at}unimo.it.
Abstract
Purpose: Several scoring systems to evaluate patients with hepatocellular carcinoma (HCC) exist. A good scoring system should provide information on prognosis and guide therapeutic decisions. The presence of variant liver estrogen receptor (ER) transcripts in the tumor has been shown to be the strongest negative predictor of survival in HCC. The aim of this study was to compare the predictive value of the commonly applied clinical scoring systems for survival of patients with HCC with that of the evaluation of ER in patients with HCC (molecular scoring system).
Materials and Methods: HCC was staged according to the Okuda classification, Barcelona Clinic Liver Cancer classification, Italian classification system (CLIP), French classification, and ER status in 96 patients. Analysis of survival was performed according to the Kaplan-Maier test and was made for each classification system and ER. A comparison between classifications was made by univariate and multivariate analysis.
Results: Among the clinical classification systems, only the CLIP was able to identify patient populations with good, intermediate, and poor prognosis. On multivariate analysis, ER classification was shown to be the best predictive classification for survival of patients with HCC (P <.0001). This difference was the result of a better allocation of patients with ominous prognosis (variant ER) having nevertheless good clinical score.
Conclusion: The evaluation of the presence of wild-type or variant ER transcripts in the tumor is the best predictor of survival in patients with HCC. Its accuracy in discriminating patients with good or unfavorable prognosis is significantly greater than that of the commonly used scoring systems for the staging of HCC.
THE SURVIVAL RATE of patients with cirrhosis and hepatocellular carcinomas (HCC) ranges from a few months to several years. This variability is related to both characteristics of the tumor and the severity of the underlying liver disease.1,2 Therefore, the classification systems in use for the staging of HCC combine information on tumor size, extrahepatic spread, and/or vascular invasion with parameters of liver function. The clinical staging systems currently used for the classification of patients with cirrhosis and HCC are the Okuda classification,3 Barcelona Clinic Liver Cancer (BCLC) classification,4 Italian classification system (CLIP),5 and French classification proposed by Chevret et al.6 The tumor-node-metastasis classification of neoplasms is only rarely used in Europe in the hepatologic environment.7 The parameters included in each scoring system are factors shown by univariate and multivariate analysis of series of patients with HCC to be predictive of survival. Although the simplicity of these scoring systems make them clinically attractive, a level of accuracy sufficient to predict survival in the individual patient cannot be reached. Therefore, the current prognostic systems are of value in predicting the average probability of survival in groups of patients but are much less accurate in predicting the probability of survival in the individual patient.
To set the prognosis and to choose the most appropriate treatment strategy is extremely important in the management of patients with HCC. Therefore, the identification of the best system to predict patient survival represents a priority for investigators in oncology. The incidence of HCC has increased during the past decades in most of the developed countries as a consequence of the epidemic of infection by hepatitis C virus (HCV).8 In fact, HCC is thought to represent the final step in the natural history of hepatitis B virus (HBV) and HCV chronic infection.9
Our group investigated the presence of estrogen receptors (ERs) in HCC. Recently, a variant form of the wild-type ER (wtER) that maintains a constitutive transcriptional activity has been described.10 The presence of this variant ER (vER) identifies HCC with a higher clinical aggressiveness, shorter doubling time as compared with the tumors characterized by wtER, and insensitivity to tamoxifen, the receptor being modified in the hormone-binding domain.11 Moreover, the presence of vER transcripts in the tumor is the strongest negative predictor of survival in inoperable HCC, and their presence is associated with a worse survival rate than that of patients with HCC expressing the wtER.12 When treated with antihormonal therapy chosen according to the presence of wtER (tamoxifen) or vER (megestrol acetate), tumor volume in all patients with wtERs halved after 9 months of tamoxifen treatment, whereas megestrol acetate was able to slow down tumor growth in patients with vERs.11 These data represented the molecular basis for a prospective, randomized study of 45 patients with HCC characterized by vERs in which it was demonstrated that megestrol acetate significantly improves survival.13
In the light of the clinical relevance of the identification of ER transcripts in patients with HCC in terms of prognosis and response to treatment, the aim of this study was to compare the prognostic value of clinical and molecular staging systems. In this study, we evaluated the survival rate of patients with cirrhosis and HCC stratified according to the stage of disease calculated using the clinical or the molecular (ER) staging systems.
MATERIALS AND METHODS
Patients
Ninety-six patients with cirrhosis and inoperable HCC consecutively seen at the Division of Gastroenterology, from 1993 to 1997 were previously studied to investigate the role of the presence of wtER and vER in the natural history of HCC.12 Because HCC in these patients was inoperable from presentation, patients received only standard supportive therapy for cirrhotic condition.
The diagnosis of HCC was made with the use of screening procedures (ultrasound and serum alpha-fetoprotein levels) and confirmed by histology. The study was conducted in accordance with the ethical standards of the Helsinki Declaration of 1975, as revised in 1983.
Etiology of Cirrhosis
Each subject was tested for HBV and HCV at entry on the study to determine the etiology of their liver disease. HBV surface antigen (HBsAg), antibody against HBV core antigen (anti-HBc), and antibody against HBsAg were determined by radioimmunoassay (Ausria II, Corab, and Axsym, respectively; Abbott Laboratories, Chicago, IL). The presence of the antibody against HCV (anti-HCV) was determined by third-generation enzyme-linked immunosorbent assay (Ortho Diagnostic Systems, Raritan, NJ).
Liver Biopsy
As part of the diagnostic evaluation, all patients underwent a liver biopsy for histopathologic diagnosis and the measurement of ER transcripts. Liver biopsies were performed with a TSK Sure-Cut 21-gauge needle (GH General Hospital, Rome, Italy) under ultrasound guidance. Two samples were obtained from the HCC. For each specimen, one biopsy core was fixed in buffered formalin and embedded in paraffin for morphologic analysis, whereas the other core was immediately stored at −80°C until it was assayed. Each patient studied was classified according to the clinical staging systems for HCC described in Tables 1 through 4⇓⇓⇓⇓.
Okuda Classification
Barcelona Clinic Liver Cancer Classification
Italian Classification System (CLIP)
French Classification
Okuda scoring system.
The Okuda classification includes the size of the tumor and parameters of liver function such as bilirubin, albumin, and the presence of portal hypertension (presence or absence of ascites). Patients with HCC are assigned to one of three stages in the Okuda staging system.
BCLC classification.
The prognostic factors included in the BCLC derive from the analysis of the factors affecting the outcome of HCC following the natural history of untreated nonsurgical tumor. Each of the four stages is obtained from the analysis of relevant prognostic variables. The most important difference between BCLC and the other scoring systems is the evaluation of the performance status test as index of the presence of cancer-related symptoms.4
CLIP classification.
Four variables are considered: the severity of liver function impairment as assessed by the Child-Pugh score,14 tumor morphology, levels of alpha-fetoprotein, and the presence or absence of intravascular dissemination. The CLIP score is arrived at by assigning a score (0, 1, or 2) to these four parameters.5
French classification.
This three-group classification system is based on the value of the prognostic score, ranging from 0 to 11, which identifies patients at low (group A, score = 0), intermediate (group B, score 1 to 5), or high (group A, score > 6) risk of death. The scoring system evaluates the Karnofsky index, serum bilirubin, alkaline phosphates, serum alpha-fetoprotein, and portal obstruction.
Molecular Classification
For evaluation of ER status, the ER transcripts, wtER and vER, were determined by reverse transcription-polymerase chain reaction for each patient as already described.10 Discrete cDNA bands were semiquantitated by digitized evaluation of their optical density after substraction of background (Un-Scan-Ittm gel, version 5.1, Silk Scientific, Orem, UT). The results were expressed as ratios of the intensity of the band of the investigated transcripts to the intensity of the band used as internal standard. The tumor was classified as wild-type or variant when either type was expressed alone or when the optical density of the band corresponding to one type was twice or more the density of the other.
The survival rate of 96 patients with HCC was determined using the classifications based on the evaluation of clinical parameters or the presence of wtER or vER. We referred to this classification of the patients with HCC as molecular or ER classification.
Statistical Analysis
The results are reported as mean ± SD or median and ranges. The predictors of survival were analyzed by the Kaplan-Meier method and compared by the Mantel log-rank test.15 A multivariate analysis was performed using the Cox regression model.16 All the constitutive variables of the clinical staging systems under evaluation and the various categories of these systems as a whole were separately analyzed. A P value less than .05 was considered to be significant. All calculations were performed using the SPSS 10.0 statistical package (SPSS, Chicago, IL).
RESULTS
As already reported, at the end of follow-up, the median duration of survival in the 96 cirrhotic patients with HCC was 22 months (95% confidence interval [CI], 17.85 to 26.15). The overall actuarial probability of survival at years 1, 2, 3, 4, 5, and 6 was, respectively, 72%, 41%, 38%, 24%, 20%, and 9% (Fig 1⇓). The clinical and demographic characteristics of the patients as well as the causes of death have been reported elsewhere.12 Briefly, 69 (71.8%) patients were male and 36 (28.1%) were female; anti-HCV was positive in 61 (63.5%) patients, whereas HBsAg was positive in 16 (16.6%). Three patients were both HBsAg- and anti-HCV–positive (3.1%), whereas eight patients (8.3%) had isolated anti-HBc and six (6.2%) had alcoholic liver disease. Eight patients (8.3%) had portal vein thrombosis at enrollment, whereas 88 did not.
Cumulative survival analysis, by Kaplan-Meier method, in the 96 patients with inoperable hepatocellular carcinoma studied.
Fifty-two patients were classified as having wtER and 44 as having vER. Patients with vERs were significantly more often HBsAg-positive than those with wtER (P = .028). This difference was more marked when, as a marker of previous HBV infection, presence of anti-HBc was considered (P = .003).
Univariate Analysis
The details of the univariate analysis by Kaplan-Meier are reported in Table 5⇓.
Univariate Analysis of the Predictive Value for Survival of the Clinical and Molecular Classifications of Hepatocellular Carcinoma
Clinical staging systems.
Okuda score and BCLC classification failed to show any significant ability to discriminate survival in our series. French and CLIP classification demonstrated good prediction of survival. However, it should be noted that in the French classification, only two patients were present in the third group, characterized by extremely poor prognosis. This may have influenced the reliability of the statistical analysis.
Molecular staging system.
The survival data were analyzed in the same group of patients according to the presence of ER transcripts in the HCC. The clinical and demographic characteristics at baseline did not differ between patients who presented wtERs or vERs, as reported previously.12 The probability of survival was higher in the 52 patients with wtER (median survival, 36 months; 95% CI, 12.92 to 59.08) as compared with that of the 44 patients with vER (median survival, 13 months; 95% CI, 8.31 to 17.69; P = .0000).
Multivariate Analysis
The constitutive variables of the clinical and molecular staging systems for HCC were individually compared by Cox regression analysis. Factors independently related to survival are shown in Table 6⇓.
Multivariate Analysis of the Predictive Value for Mortality of the Individual Variables Included in the Clinical and Molecular Classifications of Hepatocellular Carcinoma
A multivariate analysis was performed to compare the predictive value for survival of the classifications analyzed by categories for the staging of HCC considered in the study (Table 7⇓). The ER scoring system demonstrated the highest significance in terms of prediction of survival. Among the clinical classifications, only the CLIP scoring system was shown to have a statistically proven utility for the identification of patients with HCC and a good or poor prognosis. However, the level of significance was higher and 95% CI narrower for the ER classification than for the CLIP scoring system.
Multivariate Analysis of the Predictive Values for Mortality of the Clinical and Molecular Classifications of Hepatocellular Carcinoma
The patients with a good prognosis according to the Okuda, CLIP, BCLC, and French classifications were studied to determine whether this finding was confirmed by the ER classification. It was surprising to find that, for example, within CLIP stage 0 and 1 (the subgroup scores indicating best survival), the ER classification identified nine patients in stage 0 and 16 in stage 1 with vERs: these individuals had survival rates significantly below the median of the group (P = .0039). BCLC stages A and B and French group A behaved accordingly (Table 8⇓).
Survival of Patients With Cirrhosis and Hepatocellular Carcinoma (HCC) Classified as Cases With a Good Survival Rate According to the Clinical Scoring Systems and Reanalyzed According to the Type of Estrogen Receptor (ER) Transcript Present in the HCC
DISCUSSION
In this study we proposed a simple, straightforward classification for the prognostic assessment of patients with cirrhosis and HCC: the molecular or ER classification, which is based on the evaluation of a biologic characteristic of the tumor; namely, the determination of the type of ER transcripts. Our study shows that the type of ER present, wild-type or variant, sensitively identifies patients with HCC with a good or poor prognosis. In comparing the ER classification with the current clinical staging systems, it was clear that the molecular classification was more accurate in predicting the survival rate of the patients with HCC than the clinical systems in use. Thus, the ER classification represents a valuable tool for the evaluation of the patient with HCC that can be used not only for an accurate prognosis but also in clinical decision making as a reliable assessment of sensitivity to treatment.13
The evaluation of the individual components of the different classification systems shows that two of the three parameters with independent prognostic power, namely, bilirubin and portal vein thrombosis, are those common to all the clinical staging systems. This observation indicates that the additional parameters included in each of the four clinical staging systems considered have little importance in the evaluation of survival of patients with HCC. The analysis of single components shows that ER status is the strongest factor related to survival. The presence of wtER is inversely related with bad prognosis, and survival is fivefold better in the patients with HCC presenting with wtER than that of patients presenting with HCC and vER. In HCC, as with breast cancer, the onset of variant forms of ER is likely to be a function of an elevated proliferation rate of the neoplastic clone, tumor aggressiveness, and lack of hormonal control on the tumor growth rather than be a marker of cell differentiation. This comment is supported by molecular investigations aimed at determining whether a relationship between the type of ER transcript and cancer cell proliferation and local invasiveness exists in HCC. Although no difference in the percentage of apoptotic cells has been found between HCCs expressing wtER or vER, proliferation as studied by immunostaining for Ki67, PCNA, and Cyclin D1 was significantly more elevated in vER than wtER liver tumors.17 In this context, the highly significant relationship between presence of HBV infection and occurrence of vER also may be interpreted as an additional factor predisposing to abnormal proliferation: indeed, the cellular genes targeted by HBV are key regulators of cell proliferation,18 and viral HBx protein has been reported to favor early deregulated proliferation of hepatocytes during liver carcinogenesis.19
ER classification requires an invasive procedure to obtain a sample of liver tissue. This could be regarded as a flaw. However, liver biopsy is quite often performed in patients in whom HCC is suspected. In particular, it was done in approximately 70% of the patients with HCC in the studies evaluated (77.7% of the patients in the CLIP study and 63% in the French study).5,6 This observation underscores the fact that the evaluation of the ER transcripts in HCC is feasible in the clinical practice, and therefore, that the introduction of the molecular classification system can be proposed. The amount of liver tissue required for the determination of ER is extremely small. In proposing this innovative classification system for HCC, we took this observation into account. In fact, the clinician needs to perform only one liver biopsy to collect the sample needed for the morphologic analysis and the determination of ER status of the tumor, and therefore, the patient is not subjected to additional risk.
The patients with HCC with poor prognosis were recognized by almost all clinical staging systems used. It is important to recognize that CLIP, BCLC, and the French classifications claimed to be able to identify at best, patients with good prognosis. However, when HCCs belonging to the best categories of such clinical staging systems were stratified according to the presence of wtER or vER, we were able to identify within these groups a good number of HCC with vER with definitely poor survival. We concluded that their statistical weight in determining survival was diluted by the overall good prognosis of the other HCCs (wtER positive tumors) and, therefore, not perceived when examining the group as a whole.
In conclusion, the evaluation of the presence of wild-type or variant ER transcripts in the tumor is the best predictor of survival in patients with HCC. Its accuracy in discriminating patients with good or unfavorable prognosis is significantly greater than that of the commonly used clinical scoring systems for the staging of HCC.
Footnotes
-
This work was supported by grant “40% - Cofin 2000” of the Ministry of University and of Scientific and Technological Research (MURST), Progetti di ricerca con ricaduta assistenziale Azienda Ospedaliera Policlinico di Modena and Associazione Italiana per la Ricerca sul Cancro (E.V.), and Young Investigator Grant from MURST (A.C.).
- Received November 12, 2001.
- Accepted September 16, 2002.
