Staging systems are an attempt to incorporate the biology and therapy for cancer in a way that enables categorization and prediction of oncologic outcomes. Because of unusual disease biology and complexities related to treatment intervention, efforts to develop reliable staging systems for hepatic malignancies have been challenging. This article discusses the ways in which improved understanding of these diseases has informed the evolution of prognostication systems as applied to hepatocellular carcinoma, cholangiocarcinoma, and hepatic colorectal adenocarcinoma.
Key points
- •
The disease biology of primary and metastatic cancers of the liver does not always fit well into the conventional tumor-node-metastasis staging paradigm.
- •
The impact of surgical resection on long-term survival has led to the incorporation of anatomic and biological measures of resectability into prognostication systems for primary and metastatic cancers of the liver.
- •
Efforts to refine the prediction of individual oncologic outcomes have used nomograms to accommodate the reality that some prognostic factors are continuous variables, and that some prognostic factors are more important than others.
Introduction
The traditional tumor-node-metastasis (TNM) framework that underlies the staging of many epithelial malignancies is based on 2 fundamental observations regarding cancer biology. First, cancers seem to progress in a sequential manner, characterized by primary tumors progressing in size or anatomic depth of tissue involvement, lymphatic spread into regional lymph nodes, and distant hematogenous progression into discontiguous organ sites. Second, the presence of nodal metastases typically carries more prognostic weight than disease progression at the primary site, and the presence of distant metastases typically carries more prognostic impact than disease progression either at the primary site or in regional lymph nodes. The same TNM staging paradigm that has successfully enabled the development of prognostic staging systems for many cancers has never been useful for prognostic stratification of patients with hepatic malignancies. For one thing, many hepatic neoplasms are manifestations of metastatic disease; the uniform assignment of all patients into a single stage IV category does little to permit subtle distinctions of expected survival outcomes. In addition, nodal metastases (often a dominant factor in the estimation of prognosis for patients with epithelial malignancies) are often rare and occasionally of unclear prognostic significance in primary and metastatic hepatic tumors. Moreover, prognosis for patients with hepatic tumors is often determined by the technical feasibility of surgical resection; as a result, variables that describe the anatomic location (eg, involvement of critical vascular structures) or distribution of tumors (eg, the presence of bilateral disease) must be included in any meaningful prognostication scheme. In addition, many primary hepatic malignancies arise in a setting of compromised liver function, so the severity of that liver dysfunction often dictates prognosis as much as (if not more than) traditional oncologic parameters like tumor size or number. In order to accommodate this diversity and heterogeneity of prognostically informative variables, efforts to stratify expected survival for patients with hepatic malignancies have resulted in a multitude of nontraditional staging and scoring systems. This article provides an overview of the prognostication systems that have been proposed for patients with primary and metastatic hepatic malignancies.
Introduction
The traditional tumor-node-metastasis (TNM) framework that underlies the staging of many epithelial malignancies is based on 2 fundamental observations regarding cancer biology. First, cancers seem to progress in a sequential manner, characterized by primary tumors progressing in size or anatomic depth of tissue involvement, lymphatic spread into regional lymph nodes, and distant hematogenous progression into discontiguous organ sites. Second, the presence of nodal metastases typically carries more prognostic weight than disease progression at the primary site, and the presence of distant metastases typically carries more prognostic impact than disease progression either at the primary site or in regional lymph nodes. The same TNM staging paradigm that has successfully enabled the development of prognostic staging systems for many cancers has never been useful for prognostic stratification of patients with hepatic malignancies. For one thing, many hepatic neoplasms are manifestations of metastatic disease; the uniform assignment of all patients into a single stage IV category does little to permit subtle distinctions of expected survival outcomes. In addition, nodal metastases (often a dominant factor in the estimation of prognosis for patients with epithelial malignancies) are often rare and occasionally of unclear prognostic significance in primary and metastatic hepatic tumors. Moreover, prognosis for patients with hepatic tumors is often determined by the technical feasibility of surgical resection; as a result, variables that describe the anatomic location (eg, involvement of critical vascular structures) or distribution of tumors (eg, the presence of bilateral disease) must be included in any meaningful prognostication scheme. In addition, many primary hepatic malignancies arise in a setting of compromised liver function, so the severity of that liver dysfunction often dictates prognosis as much as (if not more than) traditional oncologic parameters like tumor size or number. In order to accommodate this diversity and heterogeneity of prognostically informative variables, efforts to stratify expected survival for patients with hepatic malignancies have resulted in a multitude of nontraditional staging and scoring systems. This article provides an overview of the prognostication systems that have been proposed for patients with primary and metastatic hepatic malignancies.
Primary hepatic malignancies
Hepatocellular Carcinoma
Overview
Hepatocellular carcinoma (HCC) is a leading cause of mortality worldwide. Because conditions of chronic inflammation and chronic hepatocellular death and regeneration also favor hepatocarcinogenesis, the risk of HCC is significantly higher among patients with chronic viral hepatitis and long-standing cirrhosis. In addition, these conditions effectively cause the liver to become a carcinogenic substrate, promoting the development of multifocal tumors and recurrent tumors after successful treatment. The range of treatment options for HCC include systemic chemotherapy, transarterial delivery of bland embolization particles, chemoembolization, selective internal radiation therapy, tumoral injection with cytotoxic ethanol, thermal ablation using radiofrequency or microwave energy, resection, and total hepatectomy with orthotopic allotransplantation. To a large extent, the development of such a wide breadth of treatment options has been necessary because many patients with HCC are not candidates for aggressive interventions.
Prognostic variables
Retrospective series have determined that traditional oncologic metrics like tumor size and number are reflective of disease progression and eventual prognosis. The United Network of Organ Sharing and University of California San Francisco criteria used to select patients with HCC who are most likely to benefit from liver transplantation are based on tumor size and number. The relationship between tumor size and prognosis is unlikely to be linear, because several series have reported potentially favorable outcomes among patients undergoing resection of solitary, large HCC ; the ability of tumors in these highly selected patient cohorts to reach massive proportions without showing more ominous signs of progression such as metastatic disease is probably a reflection of indolent intrinsic disease biology. The prognostic significance of regional (hilar or perihepatic) lymph node metastases remains incompletely defined; although the presence of such nodal metastases seems to be an ominous sign, the prevalence of nodal metastases seems to be low. Although the development of extrahepatic metastases portends a poor outcome, the preneoplastic nature of chronic liver diseases often makes the risk of intrahepatic recurrence higher than the risk of distant metastasis.
One prognostic variable that has consistently been shown to indicate a higher likelihood of recurrent disease after resection or transplantation is vascular invasion by tumor. Although cases of macrovascular invasion into the portal or hepatic veins can be diagnosed radiographically, the more common circumstance of microvascular invasion can be difficult to exclude before resection or transplantation. The heavy prognostic impact of microvascular invasion has led some investigators to recommend routine core biopsies of HCC tumors before transplantation. Several investigations have pointed out the possibility that other negative prognostic markers such as advanced tumor grade, marked increase of serum levels of alpha fetoprotein (AFP), or large tumor size may be surrogates of microvascular invasion.
Because of the causal relationship that exists between chronic viral hepatitis or cirrhosis and HCC, the presence of these carcinogenic conditions is a strong predictor of disease recurrence following therapy. In addition, because advanced cirrhosis limits the feasibility of aggressive liver-directed therapies, metrics like the Child-Turcotte-Pugh score or MELD (Model for End-stage Liver Disease) score that quantify hepatic dysfunction can be as important as more traditional oncologic descriptors in determining long-term prognosis and treatment morbidity and mortality. For example, although a patient with preserved liver function and 2 left-sided HCC tumors measuring 3 cm each may be eligible for potentially curative resection, a patient with advanced liver insufficiency and a solitary 2-cm HCC may be ineligible to receive life-prolonging therapy.
Several series reporting outcomes after resection of HCC have indicated that perioperative outcomes such as estimated blood loss and postoperative complications may directly affect expected survival outcomes.
Prognostication systems
Early staging systems attempted to impose the traditional TNM scheme onto HCC prognostication ( Table 1 ). In an effort to incorporate the important prognostic variables of tumor size, tumor number, and presence of vascular invasion, the 1997 American Joint Committee on Cancer (AJCC) staging system used complex combinations of these 3 variables in 10 distinct T classes. This complexity was partially ameliorated by the International Hepato-Pancreato-Biliary Association (IHPBA) staging system, which relied on a scoring system based on the absence or presence of 3 criteria (>2 cm maximal tumor dimension, multiple tumors, and vascular invasion) to define 4 distinct T classes. Both systems relied on a maximal tumor diameter cutoff of 2 cm to measure tumor size. The recognition that a cutoff of 5 cm permitted more accurate stratification of outcomes led to the Vauthey staging system, and this improvement was incorporated into the next iteration, the 2002 AJCC staging system.
| Staging System | Adverse Prognostic Variables |
|---|---|
| AJCC system, 5th edition | Tumor size >2 cm |
| Multiple tumors | |
| Bilateral tumors | |
| Microvascular invasion | |
| Major vascular invasion | |
| Extrahepatic invasion | |
| IHPBA system | Tumor size >2 cm |
| Multiple tumors | |
| Vascular invasion | |
| Vauthey system | Tumor size >5 cm |
| Multiple tumors | |
| Microvascular invasion | |
| Major vascular invasion | |
| AJCC system, 6th edition | Tumor size >5 cm |
| Multiple tumors | |
| Microvascular invasion | |
| Major vascular invasion | |
| Okuda system | Disease involving >50% liver parenchyma |
| Ascites | |
| Albumin ≤ 3 mg/dL | |
| Bilirubin ≥ 3 mg/dL | |
| BCLC system | Tumor size >3 cm |
| Tumor size >5 cm | |
| Multiple tumors | |
| Vascular invasion | |
| Extrahepatic invasion | |
| Symptoms | |
| Child-Pugh class | |
| CLIP system | Multiple tumors |
| Disease involving >50% liver parenchyma | |
| AFP ≥ 400 ng/mL | |
| Portal vein thrombosis | |
| JIS | Tumor size >2 cm |
| Multiple tumors | |
| Vascular invasion | |
| Child-Pugh score |
One potential limitation of these systems was that they did not accommodate liver dysfunction as a prognostic or treatment selection variable. An early attempt to incorporate both tumor-specific and liver function–specific variables was the Okuda staging system, which was first described in 1985. Using 4 criteria (tumor extension involving >50% of the liver parenchyma, presence of ascites, serum albumin ≤3 mg/dL, and serum bilirubin ≥3 mg/dL), this system categorizes patients into 3 stages (stage A, no criteria; B, 1–2 criteria; C, 3–4 criteria) that accurately stratify expected survival outcomes. The Cancer of the Liver Italian Program (CLIP) staging system used gradations of liver function as assessed by Child-Pugh class, tumor burden as characterized by morphologic assessments of tumor number and extent of hepatic parenchymal involvement, and the prognostic variables of AFP and portal vein thrombosis into a prognostically informative numerical scoring system that ranged from 0 to 6. The Barcelona Clinic Liver Cancer (BCLC) system, first described in 1999, also incorporated factors that described tumor burden and liver function but also introduced an overall assessment of performance status; as such, the BCLC system was developed as a means to streamline the triage of patients into appropriate treatment strategies. Stage A patients were largely patients with limited disease burden, preserved hepatic function, and excellent performance status who would be best served with liver transplantation. Stage B patients were patients with preserved hepatic function and performance status whose tumor burden exceeded criteria for transplantation; as a result, these patients were generally treated with hepatic resection. Stage C patients were those whose performance status or disease burden rendered them ineligible for resection or transplantation and who were best treated with transarterial therapies, and stage D patients were patients whose combination of liver function, performance status, and tumor burden were best managed with supportive care only. In this way, the BCLC system was designed as a means of triaging patients into appropriate treatment strategies and not strictly as a means of prognostication The merit of incorporating both traditional tumor-specific and hepatic functional metrics into prognostication for HCC was shown by the Japanese Integrated System (JIS), which relied on a combination of the IHPBA staging system and the Child-Pugh classification into a simple scoring system that stratified expected survival outcomes after resection of HCC.
In 2008, the Memorial Sloan-Kettering Cancer Center (MSKCC) experience with 184 patients with HCC treated with partial hepatectomy was examined in an effort to validate these prognostication systems. As expected, most of these systems were able to stratify overall and recurrence-free survival outcomes into distinct prognostic categories. However, none of these systems permitted accurate stratification of individual survival outcomes. In this analysis, a concordance index (c-index) was calculated for each system’s ability to properly predict the longer survivor between 2 randomly selected pairs of patients. When calculated based on every possible random pair of patients, a c-index of 1.0 indicates that a system is able to properly predict the longer survivor 100% of the time; a c-index of 0.5 indicates that a system is able to properly predict the longer survivor 50% of the time (indicating an individual predictive capacity no better than a coin toss). The c-indices for the various systems ranged between 0.54 and 0.59, and only the 2002 AJCC staging system had a 95% confidence interval that exceeded 0.5, indicating that all other staging systems were incapable of stratifying expected survival outcomes on an individual basis. In an effort to enhance individual risk prediction, the data set was used to generate a prognostic nomogram in which prognostic factors were assigned weighted scores based on the magnitude of their association with survival. In this way, a scoring system that incorporated the variables of patient age, estimated operative blood loss, margin status, presence of satellite lesions, presence of vascular invasion, tumor size, and AFP level was developed to determine the specific likelihood of overall or recurrence-free survival for individual patients. In addition to being able to allow for some variables being more prognostically informative than others, nomograms also enable the incorporation of continuous variables (eg, estimated blood loss) into risk calculation. The MSKCC HCC nomogram had a c-index of 0.74 (95% confidence interval, 0.68–0.80) in the prediction of overall survival, and a c-index of 0.67 (95% confidence interval, 0.61–0.73) in the prediction of recurrence-free survival, which indicates that the use of a prognostic nomogram can enhance the precision of individual prognostication.
Cholangiocarcinoma
Overview
Arising from the cholangiocytes populating the biliary tree, cholangiocarcinoma (CCA) can present as intrahepatic and hilar variants (with extrahepatic CCA being outside the present article’s focus on hepatic malignancies). Associations exist between CCA and hepatitis C, chronic biliary parasitic infestations, and primary sclerosing cholangitis. However, CCA differs from HCC in that most cases arise outside the context of a known predisposing hepatocellular or biliary disorder. Like HCC, CCA is optimally treated with resection; consequently, anatomic factors associated with resectability serve as prognostic indicators as well. The insidious ability of CCA to grow longitudinally along the biliary tree has led to an aggressive approach to tumoral extirpation, which is particular evident with hilar CCA, for which optimal therapy involves resection not only of the biliary tumor but of the upstream hemiliver and the contiguous biliary radicles within. In addition to longitudinal growth, CCA has the ability to grow in a radial manner, often invading from the biliary duct into the adjacent portal venous and hepatic arterial structures that make up the remainder of the portal triad. The anatomic confines of the hepatic hilus, with the close proximity of the portal veins and hepatic arteries to the hepatic ducts, means that significant bilateral involvement of critical hilar structures may render a small focus of CCA unresectable.
Prognostic variables
Like other epithelial malignancies, CCA seems to follow a familiar pattern of local tumor progression that begins with intramucosal atypia and eventually progresses to transmural invasion. Also like other epithelial malignancies, CCA has a tendency to metastasize to local hilar lymph nodes. As with HCC, tumor size and tumor number have been shown to be reliable prognostic indicators for patients undergoing surgical resection. The prevalence of hilar lymph node metastases is higher for CCA than for HCC, and the presence of nodal metastases has proved to be a powerfully negative prognostic factor. The presence of intratumoral lymphovascular invasion, which may be a surrogate marker of nodal metastases, has similarly been found to negatively affect survival outcomes. Tumor differentiation has also been shown to be associated with differences in survival. Marked increases in serum tumor markers like carbohydrate antigen 19-9 (CA19-9) and carcinoembryonic antigen (CEA) have also been associated with poorer outcomes. To an extent greater than that seen with HCC, it is common for CCA to present with distant extrahepatic metastases, and their presence clearly portends a poor prognostic outlook. As stated, CCA has an ability to grow longitudinally along biliary structures with microscopic extension beyond areas of grossly visible tumor involvement. The ability to effect a microscopically negative resection margin has been associated with improved survival outcomes.
Prognostic systems
Intrahepatic cholangiocarcinoma
The typical pattern of primary tumoral progression, nodal metastasis, and hematogenous dissemination to distant organ sites has allowed a traditional TNM approach to CCA staging ( Table 2 ). Early staging systems for intrahepatic CCA were based on parameters of tumor size and number that were similar to those used for hepatocellular carcinoma. In 2001, Okabayashi and colleagues from the National Cancer Center Hospital in Tokyo used their experience with 60 patients with resected intrahepatic CCA to identify hilar lymph node metastasis, multiple tumors, tumor-related symptoms, and microscopic or macroscopic invasion of any vascular structure as prognostic factors on multivariate analysis. Using these variables, they proposed a novel staging system in which stage I disease was defined as the presence of solitary tumors without vascular invasion, stage II disease was defined by the presence of solitary tumors with vascular invasion, stage IIIA disease was defined by the presence of multiple tumors, stage IIIB disease was defined by the presence of nodal metastases, and stage IV disease was defined by the presence of distant metastases. The ability of this staging system to stratify recurrence-free and overall survival outcomes ultimately led to the inclusion of these criteria into the current AJCC staging system, which uses a T classification system based on tumor number and presence of vascular invasion, with T3 tumors defined by tumor extension into extrahepatic tissues and T4 tumors defined by diffusely infiltrative, periductal tumor infiltration. Several investigators have since suggested that tumor number should be included as a prognostic factor. In 2013, Wang and colleagues developed a prognostic nomogram based on CEA level, CA19-9 level, vascular invasion, nodal metastases, direction of invasion into extrahepatic structures, tumor number, and maximal tumor diameter. Using methodology similar to that used by Cho and colleagues, they used their data set of 367 patients treated with partial hepatectomy to show that this prognostic nomogram outperformed the seventh edition of the AJCC staging system (with concordance indices of 0.74 vs 0.65). A similar prognostic nomogram with a concordance index of 0.69 was recently proposed by Hyder and colleagues using the prognostic variables of age, vascular invasion, nodal metastases, tumor number, maximal tumor diameter, and presence of cirrhosis using an international data set of 514 patients who underwent resection of intrahepatic CCA.
| Staging System | Adverse Prognostic Variables |
|---|---|
| UICC system | Tumor size >2 cm |
| Multiple tumors | |
| Bilateral tumors | |
| Microvascular invasion | |
| Major vascular invasion | |
| Okabayashi system | Multiple tumors |
| Vascular invasion |
Related posts:
Hepatocellular Cancer, Cholangiocarcinoma, and Metastatic Tumors of the Liver
Hepatocellular Cancer, Cholangiocarcinoma, and Metastatic Tumors of the Liver
Transarterial Chemoembolization for Primary Liver Malignancies and Colorectal Liver Metastasis
Y90 Selective Internal Radiation Therapy
Systemic Therapy for Hepatocellular Carcinoma and Cholangiocarcinoma
Imaging Approach to Hepatocellular Carcinoma, Cholangiocarcinoma, and Metastatic Colorectal Cancer
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
