Applying of pretreatment extent of disease system in patients with hepatocellular carcinoma after curative partial hepatectomy

The Pretreatment Extent of Disease System (PRETEXT) was designed for childhood liver tumors. The aim of this study was to confirm the prognostic value of the PRETEXT staging system compared with the currently and commonly used staging systems of hepatocellular carcinoma (HCC) after applying PRETEXT system in patients with HCC who underwent curative partial hepatectomy. Clinical data of consecutive patients who underwent curative partial hepatectomy were collected between February 1st, 2005 to December 30th, 2012 as the primary and internal validation cohort. Similar patients from a western hospital formed an external validation cohort. The predictive accuracy of the PRETEXT system compared with the currently used staging systems was measured by the area under the curve (AUC) on receiver operating characteristic (ROC) curve analysis. Of the 507 patients in the primary cohort, the overall median survival was 52.3 months, and the 1-year, 3-year, and 5-year overall survival rates were 83.0%, 56.8%, and 40.2%, respectively. The multivariate analysis of Cox proportional hazard regression identified INR (p=0.001), microvascular invasion (p=0.042), maximum tumor size (p=0.002) and PRETEXT staging system were independently predictors of overall survival. In the primary cohort, the AUC of the PRETEXT system was 0.702 (95% CI, 0.656 to 0.747), which was higher than the other conventional staging systems for predicting OS of HCC (P<0.01). These findings were confirmed with the internal and external validation cohorts. This study showed that the PRETEXT was a good prognostic staging system for HCC. It performed better than the conventional and commonly used staging systems in predicting survival of patients with HCC after curative partial hepatectomy.


INTRODUCTION
Hepatocellular carcinoma(HCC) is the fifth most frequently diagnosed cancer worldwide and the second most frequent cause of cancer death [1], with the highest incidence in Asian and especially in China [2]. Partial hepatectomy remains the most commonly used curative therapy modality for HCC [3,4]. Accurately prognostic prediction of HCC is important to facilitate screening of high risk patients and for the decision on adjuvant therapy. Many risk factors are associated with the prognosis of HCC which makes the tumor staging, prognosis estimation and choosing of therapy options complicated and difficult. Many clinical staging systems have been developed, taking into account tumor related characteristics, liver dysfunction, and general health status. These commonly used clinical staging systems included: (1) the 7th edition of TNM/AJCC classifica tion (TNM 7th) [5]; (2) the Barcelona Clinic Liver Cancer (BCLC) staging system [6]; (3) the International HepatoPancreatoBiliary Association (IHPBA) staging system [7]; (4) the Okuda staging system [8]; (5) the www.impactjournals.com/oncotarget Cancer of the Liver Italian Program (CLIP) staging system [9]; (6) the Groupe d'Etude et de Traitement du Carcinome He'patocellulaire (GETCH) staging system [10]; (7) the Chinese University Prognostic Index (CUPI) staging system [11]. Nevertheless, it remains controversial which of the established staging systems should be used as a universally applicable staging system to help improving the overall grim prognosis of HCC [12].
The Pretreatment Extent of Disease System (PRETEXT) was designed by the International Childhood Liver Tumor Strategy Group (SIOPEL) for staging and risk stratification of hepatoblastoma [13,14]. It was based on the anatomy of the liver and depended on the assessment of the accuracy of imaging techniques preoperatively [15]. PRETEXT system was widely used as a relatively objective method to evaluate tumor extent at diagnosis. Moreover, the system had been proved to show good prognostic value for primary malignant liver tumors of childhood [16]. Many study groups also used the PRETEXT system to describe imaging findings and perform effective comparison among different staging systems of liver tumors in children. However, no researchers applied this effectively and objectively hepatic staging system to adult liver diseases.
The aim of the present study is to apply PRETEXT staging system in predicting survival of adult patients with HCC who underwent curative partial hepatectomy. The prognostic value of the PRETEXT staging system was also compared with those obtained from the currently and commonly used staging systems of HCC mentioned above.

Characteristics of the patients
The characteristics of Eastern and Western patients enrolled in two different hepatobiliary surgery units were shown in Table 1. In these three cohorts, differences among these groups were significant for most covariates. In fact, compared to the Italian, Chinese patients were younger, with predominant hepatitis B related liver disease etiology, larger tumors and better compensated liver function, while Western patients had much higher antihepatitis C virus (HCV) positive rate.
Univariate analysis identified that gender, liver cirrhosis, Child-Pugh classification, AFP level, the international normalized ratio(INR), tumor number, maximum tumor size, microscopic vascular invasion and PRETEXT staging system were significant prognostic factors of survival after curative resection (see Table 2). The multivariate analysis of Cox proportional hazard regression identified INR (p=0.001), microvascular invasion (p=0.042), maximum tumor size (p=0.002) and PRETEXT staging system (p=0.001) were independently predictors of overall survival.

Staging systems in the three cohorts
Among the three different cohorts, patient stratification and estimated median survival time according to the 8 staging systems were depicted in Table 3. The majority of all patients were stratified to intermediate stages of the staging systems. When looking at the individual staging system as a whole, each showed a statistically significant association with prognosis. Figure 1 showed the KaplanMeier survival curves stratified according to the 8 staging systems. The detail distinction between the adjacent stages of the systems was analyzed and showed in Table 3.

Comparison of predictive performance of the PRETEXT system and other staging systems in the primary cohort
In the primary cohort, The performance of the PRETEXT system and the other seven conventional staging systems assessed by the likelihood ratio χ2, linear trend χ2 and the AIC test was described in Table 4. Compared with the other seven conventional staging systems, the PRETEXT system had better homogeneity (higher likelihood ratio χ2 score), discriminatory ability, and monotonicity of gradients (higher linear trend χ2 score). Also, it had a smaller AIC value, suggesting the predictive accuracy was higher. Moreover, the AUC of the PRETEXT system was 0.702 (95% CI, 0.656 to 0.747), which was higher than the other conventional staging systems for predicting survival of HCC (P<0.01, Figure  2

Validation of predictive performance of PRETEXT system
The results from the primary cohort were verified by two validation cohorts. The PRETEXT system had significantly larger AUC than those of other staging systems in the internal validation cohort(0.725 vs. AJCC

DISCUSSION
With the development of radiology and the different modalities of treatment for patients with HCC, increasing number of patients were detected HCC in relatively early stage [26], which we defined aspatients with no major vascular invasion, extrahepatic metastasis or lymph nodes spreading. Nevertheless, even in this selection of patients, the prognosis still varied since many predictors were related with the survival or the diseasefree survival [27,28]. Many staging systems had been proposed and they had been shown to have different ability to discriminate survival in HCC patients in different studies [29,30]. Much debate still existed regarding to which prognostic staging system was the best.On the one hand, different geographic regions were attributed to various patients characteristics which led to different predictive value of the commonly used staging systems. For example, alcohol and HCV had repeatedly been identified as two leading etiologic factors for HCC in Europe studies [31], while HBVinfection was the leading etiologic factor in Chinese patients [32,33]. On the other hand, the heterogeneity treatment options at diagnosis were associated with the lack of a consensus of different predictive value of HCC staging systems [34]. In the present study, we focused on the patients with HCC after curative partial hepatectomy, The predictive performance of the conventional staging systems were not specifically constructed for this selected group and needed to be confirmed.  After revised in 2005, PRETEXT staging system was tend to be more integral and efficient and was intended to applicable to all primary malignant liver tumor of children. It included almost all risk factors of liver tumors, such as tumor size, location, number, major vein involvement, extrahepatic abdominal disease, tumor rupture or intraperitoneal haemorrhage, distant metastases, lymph node metastases and so on [19]. More importantly, PRETEXT staging system was the only pretreatment evaluation system investigated prospectively in patients with hepatoblastoma [14]. As the widely validated effective system in predicting surgical resectability and the prognosis in liver tumor of children, we proposed to apply PRETEXT system in adult liver tumor of HCC and confirm its prognostic significance in HCC. Similarly as adult liver carcinoma TNM system used and compared with other staging systems in liver tumor of children [16,35], we proposed to assess patients with HCC using PRETEXT system and extend it to all malignant liver tumors when its accuracy and prognostic significance were well validated in different studies. Particularly necessary to point out that since this was the first study reported PRETEXT in patients with HCC, we used the findings of perioperative exploration and the pathology report of the postoperative resection specimen rather than imaging manifestations to determine the PRETEXT staging in order to ensure the accuracy of the PRETEXT staging of patients with HCC. Meanwhile, we compared preoperative findings with the golden standard of pathologic results to further confirm the accuracy of PRETEXT staging. In the present study, we showed that 78(15.4%) of 507 tumors were incorrectly staged. This might be explained by the difficulty of distinguishing tumor and nontumor tissues in the borderline of the different sections. The preoperative imaging findings might be used when the imaging quality and techniques improved or the error rate reduced to an acceptable range.
Our results showed that the PRETEXT system wasan independent risk factor for overall survival. Moreover, it showed superior prognostic value in both Eastern cohort and Western cohort compared with the commonly used staging systems. Multiple studies compared the staging systems in HCC and have reported different ranking of staging systems in predicting prognosis [3639]. In our relatively selected patients, we showed that the PRETEXT staging system performed better than the conventional staging systems in the primary cohort(p<0.05), and subsequently we validated our findings in two different cohorts involving Eastern and Western patients. Among the other seven staging systems, CLIP scores showed superior performance in predicting overall survival, whereas no significance difference was observed.
There were several limitations of this study: (1) the PRETEXT system was a comprehensive and sophisticated stage for liver tumor, while we just applied part of the PRETEXT system for a selected patients with HCC. We have planned a further study to apply whole PRETEXT system to all stages of patients with HCC; (2) originally PRETEXT system was used preoperatively to evaluate prognosis of patients, thus further studies should concentrate on apply it at diagnosis; (3) since this was the first study applying PRETEXT system for patients with HCC, more studies were needed to confirm its prognostic value for HCC.
In conclusion, this study showed that the PRETEXT was a good prognostic staging system for HCC. It performed better than the conventional and commonly used staging systems in predicting survival of patients with HCC after curative partial hepatectomy.

Patients
The inclusion criteria of patient selection were: (1) all patients submitted to hepatic resection as the initial treatment; (2) a preoperative ECOG criteria score of 01 [17]; (2) ChildPugh class A and B; (4) histologically proven HCC in the resected specimen; (5) no evidence of extrahepatic metastasis was present at the time of surgery, and at pathologic examination did not present tumor invasion into a major branch of the portal or hepatic veins, direct invasion of adjacent organs, or spread to the lymph nodes of the hepatic hilum; (6) no tumor enucleations were included in the present study and all resections considered in the present analysis were curative resections at histology. The curative resection was defined as complete resection of tumor according to the criteria that was previously reported [18]. Applying these criteria, we prospectively collected data of consecutive patients with HCC received surgery by the same surgical team between February 1 st , 2005 and December 30 th , 2012 at the Eastern Hepatobiliary Surgery Hospital (EHBH) of Shanghai, China. After excluding patients with incomplete data, the final study ultimately consisted of 507 patients as the primary cohort. In the same time period 233 patients underwent resection by another surgical team in the same hospital enrolled as internal validation cohort according to the same including criteria. Moreover, another independent cohort including 293 patients were obtained from the Department of Surgery and Transplantation of the University of Bologna during February 2000 to November 2011 and using the similar inclusion criteria to serve as the external validation cohort of this study.

PRETEXT staging system
PRETEXT staging system, which is based exclusively on imaging at diagnosis and Couinaud's system of segmentation of the liver, divides the liver into four parts, called sections ( Figure 3). The liver segments are grouped into four sections as follows: the left lobe of the liver consisting of a left lateral section (segments 2 and 3) and the medial section (segments 4); the right lobe dividing into right anterior section (segments 5 and 8) and right posterior section (segments 6 and 7). The term section is used (where other authors use segment or sector) to avoid terminological confusion. Couinaud segment 1 was not included in the original PRETEXT system. While In the 2005 revised PRETEXT system [19], tumors limited to the caudate lobe were classified as PRETEXT II. The tumor is classified into one of the following four PRETEXT categories depending on the number of liver section that are free of tumor ( Figure 3). PRETEXT I (three adjacent section free of tumor); PRETEXT II (two adjacent free of tumor); PRETEXT III (one section free of tumor or two sections in one hemiliver and one nonadjacent section in the other hemiliver) and PRETEXT IV (no tumor free section involved). There is no change in numbering for tumors involving the caudate lobe and any other part of the liver.
The assessment of the extent of the primary tumor was performed by abdominal ultrasound and computed tomography (CT). Magnetic resonance imaging or hepatic angiography was only performed if thought necessary by the surgeon of local center. In this study, patients were staged to PRETEXT system according to the findings of perioperative exploration and the pathology report postoperatively, which could ensure the accuracy of the PRETEXT staging.

Clinical staging systems
Collected data was used to restage all patients. This included all patients assessed by the TNM 7th edition, BCLC, IHPBA, Okuda, CLIP, GETCH and CUPI staging systems.

Diagnosis and treatment
After a detailed history and a complete physical examination, the hepatitis B and C serology, liver function test and tumor markers examination which included alphafetoprotein (AFP), carbohydrate antigen 199 (CA199), and carcinoembryonic antigen (CEA) was routinely performed. Other routine investigations were chest Xray, upper gastrointestinal endoscopy, abdominal ultrasound, contrastenhanced computerized tomography (CT) and/or magnetic resonance imaging (MRI). A clinical diagnosis of HCC was based on the criteria of the American Association for the Study of Liver Diseases (AASLD) [20].
The type of partial hepatectomy carried out was based on the tumor size, number, location, presence/ absence of cirrhosis and estimated volume of future liver remnant. As far as possible, anatomical liver resection was carried out basing on Couinaud's liver segments, sectors and hemilivers.
Histopathological study of the resected specimens was carried out independently by three pathologists who came to a consensus by discussion if there was any discrepancy.

Follow-up
Contrastenhanced CT scan or MRI of the abdomen was carried out once every 3 months in the first two years after surgery, and then once every 6 months thereafter. Further investigations were carried out when clinically indicated or when tumor recurrence was suspected. The diagnostic criteria for HCC recurrence were the same as used for the initial diagnosis. Overall survival (OS) was used as the primary endpoint of this study. OS was defined as the interval between partial hepatectomy and death or the last date of followup.

Statistical analysis
Continuous variables were expressed as mean ± SD (standard deviation) and compared using a two tailed unpaired Student's t test; categorical variables were compared using χ2 or Fisher analysis. Survival curves were plotted using the Kaplan-Meier method, and the logrank test was used to determine significance [21]. Factors that were deemed of potential importance on univariate analysis were included in multivariate analyses which was performed by means of the Cox proportionalhazards model using the forward logistic regression (LR) stepwise procedure for variable selection. Homogeneity (small difference in survival among patients in the same classification within each system) was determined by likelihood ratio χ2 which was generated by the Cox proportional hazards model. The discriminatory ability of each staging system (greater difference in survival among patients in different stages within each system) was measured by linear trend χ2 [22]. Additionally, To assess potential bias in comparing prognostic systems with different numbers of stages, the Akaike information criterion (AIC) value within a Cox proportional hazard regression model was used. The AIC statistic was defined by AIC = 2 log maximum likelihood + 2*number of parameters in the model. A smaller AIC value indicated a better model for predicting outcome [23,24]. The predictive performance of PRETEXT system and the other staging systems were measured using the area under ROC curve (AUC). AUCs were also used to compare the PRETEXT system and other staging systems using the Hanleyand McNeil method [25]. Statistical analyses were conducted with the SPSS for Windows version 18.0 release (SPSS, Inc., Chicago, IL) and ROC curve analysis were computed using MedCalcV.11.0.3.0 (MedCalc software, Mariakerke, Belgium). A value of P < 0.05 was considered significant in all the analysis.