The 8th Edition American Joint Committee on Cancer Staging for Hepato-pancreato-biliary Cancer: A Review and Update: Open Access

Hepatocellular carcinoma (HCC) is the most common primary malignancy in the liver, the fifth most common malignancy worldwide, and the third most common cause of cancer-related deaths globally.^4,5  Current treatment choices include hepatic resection, liver transplantation, radiofrequency ablation, and transcatheter arterial chemoembolization, yet optimal management remains controversial owing to the heterogeneity of HCC and underlying diseases attributable to different risk factors such as viral hepatitis, metabolic disorders, or toxins.^6,7  Given the considerable geographic and institutional variations, there are currently many different staging and scoring systems developed for HCC,⁸  among which the AJCC TNM staging system is widely accepted and is the most frequently used.

In the 8th edition, T1 is subdivided into T1a and T1b based on a size cutoff of 2 cm. Despite controversial opinions on the importance of vascular invasion in small HCCs,⁹  T1a includes tumor 2 cm or smaller with or without vascular invasion, whereas the presence of vascular invasion in tumor >2 cm distinguishes T2 from T1b. T3 (multiple, at least one >5 cm) was T3a in the 7th edition,¹⁰  whereas tumors involving a major branch of the portal vein or hepatic vein are upstaged from T3b in the 7th edition to T4 in the 8th edition. The N stage remains unchanged (Table 1). There is no minimal requirement for lymph node harvest for HCC, likely because the incidence of lymph node metastasis in HCC patients is low.¹¹ 

Right after the publication of the AJCC 8th edition, a study evaluated its performance on HCC using the Surveillance, Epidemiology, and End Results (SEER) database (1998–2013), and found that the 8th edition performed similarly to the 7th edition.¹²  Yet the authors suggested that the T2 category should be substratified, because survival was better for T2 solitary tumors larger than 2 cm with vascular invasion than for T2 multifocal tumors smaller than 5 cm. Further, for multifocal tumors 5 cm or smaller, those with vascular invasion had worse survival than those without, suggesting that vascular invasion and tumor size each have prognostic power.¹²  A similar study also used the SEER database (2010–2013) to assess the discriminating value of the 8th edition AJCC stage grouping, and found minimal improvement in the 8th compared with the 7th edition.¹³  Particularly, there was notable overlap in outcomes observed between stages IB (T1bN0M0)/II(T2N0M0) and IIIB (T4N0M0)/IVA (any TN1M0), and thus, a modified system combining stages IB/II and stages IIIB/IVA was proposed.¹³ 

Currently AJCC cancer staging for all solid tumors is based on anatomic measurements. Molecular information is not yet incorporated. A recent study identified differentially expressed stage-specific genes in HCC, which may enhance our understanding of the molecular determinants of HCC progression and serve as biomarkers that potentially underpin diagnosis as well as pinpointing therapeutic targets.¹⁴ 

Intrahepatic cholangiocarcinoma (ICC) is the second most common malignancy arising in liver, making up about 10% of all cholangiocarcinomas in the biliary tree.¹⁵  Treatment of ICC has recently been improved with the advent of hepatectomy and gemcitabine-based chemotherapy, but prognosis remains poor because of a high rate of recurrence.¹⁶  The construction of a reliable staging system to precisely predict the prognosis is important for developing a treatment strategy and assessing disease outcomes. ICC was previously staged the same as HCC in the AJCC staging manual until the 7th edition, where T category was determined by tumor numbers, vascular invasion, and invasion to adjacent organs. The 8th edition includes several significant modifications, particularly in the T category, where a size cutoff of 5 cm is set up to divide T1 into 1a (≤5 cm) and 1b (>5 cm). Previous T2a (solitary, with vascular invasion) and 2b (multiple, with or without vascular invasion) are combined as T2 to reflect the equivalent prognostic value of vascular invasion and tumor multifocality. Local direct invasion into extrahepatic structure, previously T3, is now upgraded to T4, and periductal invasion is removed from the T4 category. The N stage remains unchanged. For stage grouping, the previous T4 tumors and regional lymph node metastasis (N1) are downgraded from stage IV to IIIB, with stage IV designated only when there is distant metastasis (M1) (Table 2).

The prognostic performance of the 8th edition was overall not markedly improved over the 7th edition based on several validation studies, and potential modifications have been proposed.^17–20  Interestingly, although one study from a Korean group concluded that the new T3 did not provide good prognostic contrast,¹⁷  a much larger study across 14 major hepatobiliary centers worldwide found the new T3 and stage III to stratify the risk of death for ICC patients.²⁰  On the other hand, a Japanese group found significant overlaps between T2, T3, and T4 on 5-year disease-specific survival rate, and the authors suggested modifying the size cutoff point at 2 cm for T1 (T1, size ≤2 cm without other factors; T2, size >2 cm without other factors), and keeping periductal infiltrating in T4 because of its prognostic significance.²¹  Similarly, a Chinese study using a multicentric cohort found visceral peritoneum invasion not an independent risk factor and modified the 8th edition staging system by removing this factor, combining T1b into T2, and incorporating serum tumor markers (CA19-9 and CEA) into stage classification.¹⁹  The authors claimed that their modifications had better discriminatory capacity and validated it through a SEER cohort.¹⁹  It was not the first time incorporating serum tumor markers for staging ICC has been proposed.²²  Likewise, tumor budding at the invasive front as a known risk factor in colorectal carcinoma has recently been investigated in ICC and other tumors of the biliary tree,^23–25  which may be incorporated in the future AJCC edition for ICC.

Cholangiocarcinoma develops anywhere within the biliary tree from the most proximal intrahepatic bile duct to the most distal intraduodenal bile duct. Extrahepatic cholangiocarcinoma was separated into perihilar cholangiocarcinoma (PHC) and distal cholangiocarcinoma because of the distinct characteristics in pathology, treatment, and prognosis.²⁶  PHC, also known as Klatskin tumor, arises in the area of the biliary duct bifurcation proximal to the cystic duct and accounts for 60% to 70% of all cholangiocarcinomas.²⁷  Because of its unique local and regional growth patterns, it is staged predominantly according to tumor relationship with surrounding tissues in both the 7th and 8th editions of the AJCC staging system. Major changes in the 8th edition include incorporation of high-grade biliary intraepithelial neoplasia into carcinoma in situ (Tis) and removal of bilateral second-order biliary radical invasion from T4. Based on a large 8-institution study from Japan,²⁸  a unified staging theme for the N category in the pancreatobiliary system was adopted, in which N1 is designated as metastasis in 1 to 3 lymph nodes and N2 as metastasis in more than 3 lymph nodes. Accordingly, the stage group for T4 tumors is downgraded from stage IVA to stage IIIB, the N1 category is upgraded from IIIB to IIIC, and N2 is classified as stage IVA (Table 3). There is currently no requirement for a minimal number of lymph nodes to be evaluated for PHC.

The performance of the newly released 8th edition AJCC staging system in PHC, particularly the modified T category, is still unsatisfactory based on 2 European studies, 1 from the Netherlands and the other from Italy.^29,30  Further refinements were suggested. One proposal is to follow the same staging algorithm as that for the distal common bile duct, where a depth-based rather than layer-based measurement is applied to define T stage.³¹  Using a 4-tier invasive tumor thickness with cutoff points of 1, 5, and 8 mm, these new defining parameters appear to be an adequate alternative, improving the concordance index from 0.589 to 0.598.³¹  Other proposals include incorporating more biological factors such as microvascular invasion, perineural invasion, and tumor differentiation to ensure a personalized approach for prognostication and treatment.³²  In contrast to the T category, the new N category in the 8th edition appeared to perform better than that in the previous edition.³⁰  Besides the total number of positive lymph nodes, the parameters of lymph node ratio or log odds of metastatic lymph nodes were also proposed to improve predicting power.³³ 

Distal bile duct cholangiocarcinomas (DCCs), located between the cystic duct entry and the end of the common bile duct in the ampullary region, account for 20% to 30% of all bile duct carcinomas. The incidence rate varies considerably owing to varying prevalence of risk factors such as stone disease, parasitic infection, abnormal junction, and ulcerative colitis. DCCs are usually mass forming, displaying variable growth patterns. Traditionally, bile duct tumors were grouped as proximal, middle, and distal, but staged as a single entity with one TNM classification. The AJCC 7th edition started to separate PHC and DCC, but using a common anatomic layer–based staging system, which was described as vague and resulting in wide interobserver variations.^34,35  The 8th edition kept the layer-based approach for PHC, while adopting a more subjective depth-based approach for DCC, in that the DOI with cutoff values of 5 and 12 mm defines the T category (T1, <5 mm; T2, 5–12 mm; T3, >12 mm). Tumor infiltration into adjacent organs such as the pancreas, duodenum, and gallbladder (previously T3) is excluded from staging. Similar to PHC, Tis includes high-grade biliary intraepithelial neoplasia, and the N category is unified based on the number of lymph node metastases (N1, 1–3; N2, >3; Table 4). A minimum number of 12 lymph nodes examined for accurate staging has not been determined but has been suggested. Other changes include adding high-grade neuroendocrine carcinoma for consistency with other gastrointestinal and hepatobiliary designators and updated histologic types to match current World Health Organization terminology.

A Korean study³⁶  evaluated those changes in the AJCC 8th edition staging system and concluded that both the new T and N categories accurately predict patient prognosis in a way superior to the 7th edition. However, another study from Korea³⁷  found little survival difference using the new T criteria, and the authors believed that organ invasion should be still included as one of the factors that determine the AJCC stage given that those with dual-organ invasion had a shorter survival time than those with single-organ invasion. Nevertheless, DOI is a validated approach for categorizing T stages in many other organ systems, including the distal bile duct. A potential modifiable measurement is to find the best cutoff value for DOI. For example, revising the DOI cutoff points to 1, 5, and 10 mm was validated as a better T system,³⁸  whereas DOI cutoff points at 3 mm or smaller, 3–10 mm, and 10 mm or larger seemed to perform better in another multi-institutional study.³⁵  The superior N category in the 8th edition was validated in another study with a proposal of a minimum of 12 regional lymph nodes to be retrieved for accurate staging.³⁹ 

Gallbladder cancer (GBC) is the most common biliary tract malignancy, relatively rare in Western countries but a substantial health issue in certain regions of the world.⁴⁰  It usually presents at a late stage, as there are no specific symptoms in its early stages. The 5-year survival rate is estimated to be 5%.^40,41  Compared with other HPB cancers, the 8th edition AJCC cancer staging system for GBC did not change dramatically, but continued to use the layer-based approach for T classification, with slight changes for T2, where tumors involving the perimuscular tissue on the peritoneal side are T2a and those on the hepatic side are T2b, as the latter were found to be associated with lower survival rates.^41,42  For N category, the 8th edition adopted the unified number-based approach,⁴³  where N1 is designated as metastasis in 1 to 3 lymph nodes and N2 as metastasis in 4 or more lymph nodes (Table 5). Metastases to celiac, superior mesenteric, and peripancreatic lymph nodes, previously N2 in the 7th edition, are now considered distant metastasis (M1) in the 8th edition. To avoid underestimation of disease stage, especially in node-positive patients, a minimum of 6 lymph nodes to be harvested for histologic evaluation is recommended.

The new GBC staging system was validated by a study performed on a National Cancer Institute database, confirming that it offers adequate discrimination, especially for node-positive patients.⁴⁴  Yet, there are mixed opinions in other studies, especially for the T category.^45,46  Some proposed potential modifications including further dividing T3 into T3a (tumors penetrating serosa but not directly invading liver and/or an adjacent organ or structure) and T3b (tumor penetrating serosa and directly invading liver and/or an adjacent organ or structure).⁴⁵  The number-based N category appears to be prognostic if 6 or more lymph nodes are available for examination. Evidence that patients without nodal dissection had significantly poorer survival than those with N0 supports the 8th edition recommendation of a minimum of 6 lymph nodes to be examined. The regrouping for tumors metastatic to celiac, superior mesenteric, and peripancreatic lymph nodes as M1 (stage IVB) was challenged by a study showing that those patients with R0 resection had longer survival than patients with true distant metastatic (M1) diseases.⁴⁷ 

The ampulla of Vater is a complex structure referring to the confluence of the distal common bile duct and the main pancreatic duct in the second portion of the duodenum, or the termination of the common bile duct if the pancreatic duct enters the duodenum separately. Although ampullary carcinoma (AC) is the most common small intestinal malignancy, its incidence is rare, representing less than 0.5% of all gastrointestinal cancers, much less common than carcinoma of the pancreas or bile ducts.⁴⁸  Ampullary carcinoma may arise within ampulla (intra-ampullary) or in periampullary duodenum, and sometimes can be difficult to differentiate from periampullary carcinomas of pancreas, common bile duct, or duodenum. It is also highly challenging to stage AC because of its rarity, marked anatomic complexity, and 3-dimensional spread of the tumors in this area.⁴⁹  Since 2010, AC has been staged in the 7th edition AJCC/Union for International Cancer Control system using a “sphere model” where the ampulla is presumed to be sequentially covered by duodenum, pancreas, and then peripancreatic soft tissues, so that the stage increases when tumor invades further in that hypothetical order.⁵⁰  This model oversimplified the structural complexity and has been proved to be irreproducible and lack clinical relevance. The 8th edition modified both T and N classifications, where T stage is divided based on DOI into the duodenum (T1b, invasion into the duodenal submucosa, versus T2, invasion into the duodenal muscularis propria) and pancreas (T3a, invasion into the pancreas ≤0.5 cm, versus T3b, invasion into the pancreas >0.5 cm). T4 is limited to tumors involving the celiac axis, superior mesenteric artery, and/or common hepatic artery. Similar to pancreas and biliary duct carcinomas, the N classification is based on positive lymph nodes (N1, 1–3; N2, >3), and a minimum number of 12 lymph nodes in Whipple (pancreaticoduodenectomy) resection is recommended for optimal staging (Table 6).

Two studies have since investigated the clinical relevance of the 8th edition AJCC staging system on AC, both showing satisfactory results in the N category but not the T category. Specifically, Imamura et al⁵¹  analyzed 104 consecutive patients from a single institution and found that the T classification failed to discriminate T1b (invading duodenal submucosa) and T2 (invading duodenal muscularis) tumors, yet worked to separate T3a (invading pancreas ≤0.5 cm) from T3b (invading pancreas >0.5 cm). Therefore, the authors suggested reclassifying T1b into T2. Kim et al⁵²  analyzed 369 operatively resected AC patients and also showed unsatisfactory stratification for T staging. However, they proposed a different method of modification by eliminating the current subcategories, that is, merging T1a/T1b as new T1, keeping T2, and merging T3a/T3b as new T3. The discrepancies of results leading to different proposals call attention to the challenges of validating staging systems based on retrospective single-institution series. The unique constraints that pertain specifically to AC, which is not uncommonly misdiagnosed as pancreatic ductal adenocarcinoma (PDAC), distal common bile duct carcinoma, or duodenal adenocarcinoma, also contributes to different conclusions. Further, AC can be subdivided into intestinal or pancreatobiliary subtypes based on histomolecular profiling, which contributes significantly to different prognosis and treatment responses.⁵³  Lack of subtype data in studies trying to validate the AJCC staging system may also potentially bias the results. Overall, there is still much work to do before a reliable staging system on AC is available.

Despite tremendous scientific efforts and advancing knowledge in cancer biology, PDAC remains an aggressive malignancy prone to metastasize and difficult to treat. The 5-year survival rate is less than 10%, and all patients are eventually expected to die from the disease.⁵⁴  Accurate tumor staging is a prerequisite for further treatment and prognostic prediction. The AJCC T stage for pancreatic cancer originally applied a measurement of tumor extension (limited to pancreas versus extending to peripancreatic soft tissue) in its previous editions. After decades of debate on the definition of invasion into peripancreatic soft tissue and its reliability in survival prediction,⁴⁹  the 8th edition finally introduced a completely different size-based T staging system, where stages T1 through T3 are redefined purely by tumor size (T1, ≤2 cm; T2, >2 cm and ≤4 cm; T3, >4 cm), with T1 further divided into 1a (≤ 0.5 cm), 1b (>0.5 and <1 cm), and 1c (1–2 cm) to encompass small invasive carcinoma increasingly detected in association with cystic neoplasms such as intraductal papillary mucinous neoplasms (IPMNs), intraductal tubulopapillary neoplasms, and mucinous cystic neoplasms.⁴⁹  Whether tumor is confined to pancreas or invading into peripancreatic soft tissue does not affect staging. T4 is defined when tumor involves large blood vessels (celiac axis, common hepatic artery, and/or superior mesenteric artery). Resectability is removed from staging purpose because of its subjectivity and more tumors becoming resectable. As for PHC, DCC, GBC, and AC, the N classification for PDAC is further subdivided into N0, N1 (1–3) and N2 (>3). T1-3N2M0 is classified as stage III, whereas the other stages remain unchanged (Table 7). Microscopic evaluation of at least 12 lymph nodes is also recommended for Whipple resections. Lastly, the AJCC 8th edition clearly documents the following surgical resection margins to be evaluated: pancreatic neck/parenchymal, uncinate (retroperitoneal/superior mesenteric artery), bile duct, proximal (gastric or duodenal), and distal (duodenal or jejunal), with a new rule stating that “the presence of tumor at or within 1 mm of resection constitutes a positive margin.” Although controversial, there is no specification which margin(s) should meet the 1 mm requirement, and it is interpreted that this rule applies to all true resection margins.^55,56 

The 8th edition of the TNM staging system in exocrine pancreas, particularly PDAC, is generally well received and considered superior to the 7th edition, with more even distribution among stages and more powerful discrimination.^57–62  Yet slightly varied opinions and some refinements have been brought up in different studies. Two studies,^60,61  one from China and one from Germany, found that the new T classification had a significant advantage over the previous edition in predicting overall survival whereas the new N classification did not, which is opposite to the results of an international multicenter cohort study⁶²  where the revised N stage was highly prognostic, but the revised T stage was not. A Chinese study⁶³  using the SEER database proposed to combine stage IIA with IIB and further subclassify stage III for better discriminative power. The latter proposal was endorsed by authors of another study⁶⁴  using the SEER database, who suggested subclassifying stage III into IIIA (T[1–3]N2/T4N[0–1]) and IIIB (T4N2) because of significant differences in overall and disease-specific survival. For node-positive patients, some studies again advised using positive lymph node ratio as a better nodal parameter compared with the total number of positive lymph nodes.⁶⁵ 

The validation of the 8th edition staging system was done mostly on PDAC, yet one study investigated on IPMN with associated invasive carcinoma.⁶⁶  Using the SEER database, the authors found that tumor size of invasive IMPN was not a predictor of survival in patients with a resectable tumor larger than 2 cm (size >4 cm versus >2 and ≤4 cm), and that the 7th edition was more applicable than the 8th edition staging classification. However, a major caveat is how tumor size was defined in those studies: did it include the whole tumor with IMPN component, or was only the size of the invasive component included? Further studies may be warranted with a clear definition of tumor size and then comparison with the 7th edition to validate IPMN or mucinous cystic neoplasm with an associated invasive carcinoma.

The new size-based tumor classification and the new 1 mm or smaller positive margin rule brought up considerable controversies on how to best gross a Whipple specimen to yield best information for staging, particularly tumor origin, size, margin status, and lymph node yields.⁶⁷  The 2 most commonly used grossing protocols are axial sectioning^68,69  and bivalving methods,^70,71  each having pros and cons. The axial sectioning method is more accurate for tumor size and margin status, but may lead to repeated lymph node counting, and is not suitable for AC or IPMN. The bivalving method is beneficial in evaluating AC or IPMN, with a better lymph node yield and more accurate positive lymph node ratio, but is not ideal for margin status. Which methods to use may also depend on clinical setting; for example, cancer status post chemotherapy is generally difficult to resect and an axial sectioning method may be preferred in order to accurately evaluate margin status in such cases. Thus, standardization of Whipple grossing protocol may not be possible or necessary, and different techniques should be applied on a case-by-case basis. Knowing the complete clinical history, radiographical findings, and any presurgical treatment is critical to decide which technique to use so as to yield the most helpful information for tumor staging in the proper clinical context. For studies validating the changes of AJCC staging, the Whipple grossing methods may be taken into account for data analysis in order to reach a comparable and reliable conclusion.

Neuroendocrine tumors (NETs) of the hepatobiliary system are extremely rare. Currently no separate staging system is available for NETs of liver, bile ducts, or gallbladder. However, NET is relatively common in the pancreas, accounting for approximately 1% to 2% of all pancreatic tumors with unique pathologic features and clinical behavior. The World Health Organization classified pancreatic NETs into 3 grade groups according to mitotic count and Ki-67 labeling index. Pancreatic NETs were originally staged the same as exocrine cancers in the AJCC/TNM system. The European Neuroendocrine Tumor Society was the first to introduce a size-based staging parameter for pancreatic NETs, which is now incorporated and modified into the AJCC 8th edition (Table 7). Since its release, several studies have validated and supported the clinical use of this new staging system over the 7th edition or the European Neuroendocrine Tumor Society system.^72,73 

There have been major advancements in our understanding of cancer biology and discovery of new factors that predict cancer outcomes and treatment responses, which have led to some of the radical changes in the 8th edition AJCC cancer staging manual, particularly in HPB cancers. This new edition of the staging system has been validated worldwide. Some changes appear to perform better, and some do not. Generally, for the T category, a more objective method (tumor size, DOI, etc) appears to work better, particularly for DCC and PDAC, and a more subjective anatomic layer–based approach, such as for AC and GBC, remains suboptimal. The unified number-based N category for pancreatobiliary cancers and ACs is well received and is considered superior to its previous versions. It also rationalizes a minimal number of lymph nodes to be evaluated, although this requirement is not recommended for all HPB cancers yet. Lastly, although most of the publications on validating those changes/updates are of good quality, larger studies with solid databases are still needed to solve the remaining controversial issues in the 8th edition.

In summary, the TNM-based AJCC cancer staging system has been widely accepted and used for solid tumors; however, in an era of personalized medicine some nonanatomic factors, such as serum and molecular biomarkers, have been gaining attention for potential inclusions into the staging system. Because the AJCC promulgates staging practices through each new edition in an effort to provide a powerful, knowledge-based resource for the battle against cancer, this review summarizes the major changes of the 8th edition on HPB cancers and reviews literature validating those changes, hoping that this information will provide the reader with a better understanding of the rationales behind the changes and potential exciting developments in future editions.