Proposal for a new T-stage classification system for distal cholangiocarcinoma: a 10-institution study from the U.S. Extrahepatic Biliary Malignancy Consortium

Abstract

Background

Seventh AJCC distal cholangiocarcinoma T-stage classification inadequately separates patients by survival. This retrospective study aimed to define a novel T-stage system to better stratify patients after resection.

Methods

Curative-intent pancreaticoduodenectomies for distal cholangiocarcinoma (1/2000-5/2015) at 10 US institutions were included. Relationships between tumor characteristics and overall survival (OS) were assessed and incorporated into a novel T-stage classification.

Results

176 patients (median follow-up: 24mo) were included. Current AJCC T-stage was not associated with OS (T1: 23mo, T2: 20mo, T3: 25mo, T4: 12mo; p = 0.355). Tumor size ≥3 cm and presence of lymphovascular invasion (LVI) were associated with decreased OS on univariate and multivariable analyses. Patients were stratified into 3 groups [T1: size <3 cm and (−)LVI (n = 69; 39.2%); T2: size ≥3 cm and (−)LVI or size <3 cm and (+)LVI (n = 82; 46.6%); and T3: size ≥3 cm and (+)LVI (n = 25; 14.2%)]. Each progressive proposed T-stage was associated with decreased median OS (T1: 35mo; T2: 20mo; T3: 8mo; p = 0.002).

Conclusion

Current AJCC distal cholangiocarcinoma T-stage does not adequately stratify patients by survival. This proposed T-stage classification, based on tumor size and LVI, better differentiates patient outcomes after resection and could be considered for incorporation into the next AJCC distal cholangiocarcinoma staging system.

Introduction

Extrahepatic biliary malignancies are a combination of rare tumors (gallbladder cancer, hilar cholangiocarcinoma, and distal cholangiocarcinoma) with an estimated incidence of 11 420 patients per year in the United States.¹ Within this group, distal cholangiocarcinoma is defined by its location along the common bile duct and represents 29–42% of resected cholangiocarcinoma.^{2, 3} While chemotherapy and chemoradiation may offer improved survival outcomes in select patients, the only potential curative therapy for distal cholangiocarcinoma is resection.^{4, 5, 6, 7, 8, 9} Even after resection, median overall survival is variable ranging from 22 to 42 months.^{2, 9, 10, 11, 12}

This wide range of outcomes reflects the broad spectrum of this disease. As such, accurate staging is key, as it plays an integral role in providing prognostic information to patients and physicians. Additionally, stage can guide decision-making for administration of adjuvant therapy or clinical trial enrollment. In the current 7th American Joint Committee on Cancer (AJCC) staging system for distal cholangiocarcinoma, T-stage is defined solely by extent of tumor invasion (Table 1).¹³ The applicability of this system has been criticized, however, as the histologic landmarks defining the bile duct wall are often difficult to appreciate on pathologic review, and there remains an ambiguity in assigning T-stage to tumors expanding beyond the wall of the intrapancreatic bile duct.^{14, 15} Additionally, in studies evaluating prognostic factors in distal cholangiocarcinoma, increasing T-stage has not been associated with decreased survival.^{4, 10, 16} In fact, in a recent study where Wiltberger et al. sought to validate this 7th AJCC staging system, overall stage was associated with survival, yet progressive T-stage in lymph node negative disease was not associated with prognosis.¹⁷

Table 1.

Current American Joint Committee on cancer T-stage classification system for distal cholangiocarcinoma

T-stage	Description
T1	Tumor does not invade beyond the bile duct wall
T2	Tumor invades beyond the bile duct wall but does not invade adjacent organs
T3	Tumor invades adjacent organs but does not invade the celiac or superior mesenteric artery
T4	Tumor invades the celiac or superior mesenteric artery

Adapted from 7th Edition AJCC Cancer Staging Manual guidelines.¹³

Thus, the aim of this study was to create a novel pathologic T-stage classification system for resectable distal cholangiocarcinoma that incorporates tumor-specific factors to better predict overall survival of these patients.

Methods

Patient population and data collection

The United States (US) Extrahepatic Biliary Malignancy Consortium consists of 10 academic surgical centers: Emory University, The Johns Hopkins University, Stanford University, Vanderbilt University, Washington University in St Louis, University of Wisconsin, University of Louisville, Wake Forest University, The Ohio State University, and New York University. All patients who underwent operative interventions for extrahepatic biliary malignancies were retrospectively identified. Clinicopathologic and treatment data were collected through review of the clinical chart. Survival was determined primarily from the medical record supplemented by data from the Social Security Death Index Database. This study was approved by the institutional review boards of each participating center.

From this multi-institutional database, patients who underwent curative-intent pancreaticoduodenectomy without liver resection for invasive distal cholangiocarcinoma were included. Non-curative-intent procedures were defined as those with gross residual disease remaining after the operative intervention and included palliative procedures, explorations that were positive for unresectable or metastatic disease, and resections with grossly positive margins (R2). Patients with mid-bile duct tumors treated with bile duct resection alone were not included. Patients without sufficient pathologic data on lymph nodes, tumor size, and lymphovascular invasion (LVI) as well as those who died within 30 days of surgery were excluded.

Pathologic staging was assigned based on the 7th AJCC pathologic staging guidelines.¹³ Overall survival was defined as the percentage of patients who did not die during the time period from pancreaticoduodenectomy until the last clinical follow up date. LVI status and tumor size were determined from pathologic assessments.

Statistics

Statistical analyses were conducted using SPSS Statistics 23.0 (Armonk, NY: IBM Corp.). A p-value <0.050 was considered statistically significant. Univariate Cox regression was used to assess tumor-specific factors associated with overall survival. Tumor size was analyzed at 1 cm increments (1 cm–5 cm). Variables with statistically significant associations with overall survival on univariate analysis were included in multivariable Cox regression models. Kaplan–Meier long-rank tests were used to compare overall survival in the current and proposed staging systems.

Results

Patient population, treatment, and pathology

Of 1092 patients who underwent operative interventions for extrahepatic biliary malignancies, 295 (27.0%) patients had distal cholangiocarcinoma. Two hundred eight (19.0%) of these patients underwent curative-intent pancreaticoduodenectomy. Eight patients who died within 30 days of surgery and 24 patients with insufficient pathology data were excluded. A total of 176 (16.1%) patients were included and are detailed in Table 2. Of these, the median age was 66 years (range: 33–87), and 108 (61.4%) were male. Eight patients (4.5%) underwent portal vein resection at the time of pancreaticoduodenectomy. The median tumor size was 2.1 cm (range 0.1–7.5 cm), and 42 patients (23.9%) had tumors larger than 3 cm. LVI was present in approximately half of the resected tumors (n = 90, 51.1%). Per the 7th AJCC T-stage classification system, most patients had tumors that were T3 (n = 125, 71.0%; Table 2). Nodal metastases were present in 98 (55.7%) patients. While few patients received neoadjuvant chemotherapy (n = 2, 1.1%) or radiation therapy (n = 1, 0.6%), adjuvant chemotherapy (n = 107, 60.8%) and radiation therapy (n = 77, 43.8%) were common (Table 2).

Table 2.

Patient characteristics, treatment, and pathology of resectable distal cholangiocarcinoma

Variable	Patients analyzeda (n)	n (%) or median (range)
Male	176	108 (61.4)
Age (years)	176	66 (33–87)
BMI (kg/m2)	142	27 (16–70)
Preoperative clinical jaundice	175	129 (73.3)
Resection	176
Pylorus-preserving pancreaticoduodenectomy		86 (48.9)
Classic pancreaticoduodenectomy		90 (51.1)
Portal vein resection	176	8 (4.5)
Tumor size (cm)	176	2.1 (0.1–7.5)
Tumor size ≥1 cm	176	161 (91.5)
Tumor size ≥2 cm	176	107 (60.8)
Tumor size ≥3 cm	176	42 (23.9)
Tumor size ≥4 cm	176	11 (6.3)
Tumor size ≥5 cm	176	7 (4.0)
Lymphovascular invasion	176	90 (51.1)
Perineural invasion	174	140 (79.5)
High grade	173	49 (27.8)
R1 resection margin	176	36 (20.5)
AJCC T-stage	174
1		8 (4.5)
2		32 (18.2)
3		125 (71.0)
4		9 (5.1)
Lymph node metastasis	176	98 (55.7)
Neoadjuvant chemotherapy	176	2 (1.1)
Neoadjuvant radiation therapy	176	1 (0.6)
Adjuvant chemotherapy	163	107 (60.8)
Adjuvant radiation therapy	161	77 (43.8)

^aPatients with available data; BMI, body mass index; R1, microscopically negative margins; AJCC, American Joint Committee on Cancer.

Survival

The median follow up for survivors was 24 months (IQR: 9–42 months). Death occurred in 111 patients (63.1%) during the follow-up period. The median overall survival for all patients was 24.5 months. On univariate analysis of tumor-specific factors, current AJCC T-stage was not associated with overall survival (Fig. 1A; Table 3). Yet, tumor size ≥3 cm and presence of LVI were associated with decreased overall survival on both univariate and multivariable analyses (Table 3).

Figure 1.

Kaplan–Meier curve of overall survival after curative-intent pancreaticoduodenectomy for distal cholangiocarcinoma as related to T-stage. A) Current American Joint Committee on Cancer T-stage classification system. B) Novel proposed T-stage classification system

Table 3.

Cox regression analysis of tumor characteristics associated with decreased overall survival in resectable distal cholangiocarcinoma

	Univariable analysis				Multivariable analysis**
Variable	n	HR	95% CI	p-Value	HR	95% CI	p-Value
Tumor size ≥1 cm	176	1.26	0.61–2.60	0.527
Tumor size ≥2 cm	176	1.35	0.91–2.00	0.136
Tumor size ≥3 cm	176	1.56	1.03–2.37	0.035*	1.59	1.05–2.41	0.029*
Tumor size ≥4 cm	176	0.93	0.43–2.01	0.859
Tumor size ≥5 cm	176	1.29	0.57–2.94	0.547
Lymphovascular invasion	176	1.60	1.10–3.34	0.014*	1.62	1.11–2.37	0.012*
Perineural invasion	174	1.32	0.79–2.21	0.295
High grade	173	1.24	0.81–1.86	0.309
T-stage	174
AJCC T2 (Ref T1)		1.45	0.50–4.24	0.495
AJCC T3 (Ref T2)		0.87	0.51–1.33	0.434
AJCC T3 (Ref T4)		1.82	0.88–3.79	0.107

n, number of patients included in each univariate analysis; HR, Hazard ratio; CI, confidence interval; AJCC, American Joint Committee on Cancer; *p < 0.05; **176 patients included in multivariable analysis.

Proposed novel T-stage classification system

Patients were stratified into 3 groups by tumor size and LVI criteria to create a novel T-stage classification system (Table 4A). T1 was defined as tumor size <3 cm with absence of LVI (n = 69; 39.2%); T2 included tumor size ≥3 cm with absence of LVI or tumor size <3 cm and presence of LVI (n = 82; 46.6%); and T3 contained tumor size ≥3 cm with presence of LVI (n = 25; 14.2%). Each progressive T-stage on the new proposed system was associated with decreased overall survival (Fig. 1B). Additionally, as the proposed T-stage increased, the frequency of perineural invasion, positive resection margins, and lymph node metastases increased (Table 4B). In regards to treatment, patients with higher proposed T-stages (T2 and T3) were more likely to have received adjuvant radiation therapy with similar rates of receipt of adjuvant chemotherapy compared to T1 patients (Table 4B).

Table 4A.

Novel proposed T-stage classification system for resectable distal cholangiocarcinoma

T-stage	Description
T1	Tumor size <3 cm and LVI negative
T2	Tumor size <3 cm and LVI positive or tumor size ≥3 cm and LVI negative
T3	Tumor size ≥3 cm and LVI positive

LVI, lymphovascular invasion.

In subgroup analyses of patients with node-positive disease, this proposed T-stage system persisted in being associated with decreased median overall survival (T1: n = 24, 38 mo; T2: n = 55, 22 mo; T3: n = 19, 8 mo; p = 0.005). In patients with node-negative disease, although this novel T-stage system was not statistically associated with overall survival, there was a clear trend on Kaplan–Meier log-rank analysis (T1: n = 45, 38 mo; T2: n = 27, 22 mo; T3: n = 6, 8 mo; p = 0.617).

Discussion

In this multi-institutional study of the US Extrahepatic Biliary Consortium, outcomes of 176 patients who underwent curative–intent pancreaticoduodenectomy for distal cholangiocarcinoma were assessed. In this population, as in other cohorts, the current AJCC T-stage system was not predictive of survival; however, tumor size of ≥3 cm and presence of LVI were found to be independently associated with decreased overall survival. Thus, tumor size and LVI were incorporated into a novel pathologic T-stage system for resectable disease. Each progressive T-stage of this proposed classification was associated with a decrease in survival. To the authors' knowledge, this represents the largest study to date to evaluate the current AJCC T-stage system and to propose a novel, more predictive T-stage classification in patients who have undergone resection.

Staging for distal cholangiocarcinoma was first described in the 3rd AJCC Staging Manual.^{17, 18} In this historic classification, distal cholangiocarcinoma and hilar cholangiocarcinoma were grouped and staged together as extrahepatic bile duct malignancies, a grouping which persisted through the 6th Edition of AJCC staging.¹⁹ However, in 2009, with advent of the current system (7th), distal cholangiocarcinoma was given a unique classification, distinct from hilar malignancies. T-stage in this current system was defined solely by extent of tumor invasion.¹³

Yet, application of these criteria to resected tumors from a histologic perspective has been challenging. The distinction between T1 and T2 tumors is currently defined as invasion beyond the bile duct wall; however, for many of these tumors, the bile duct wall may be difficult to appreciate microscopically given the presence of inflammation and variation of normal histology typically associated with these malignancies.^{14, 15} Additionally, while T2 tumors invade beyond the bile duct wall, T3 tumors invade adjacent organs, such as the pancreas. Thus, for tumors invading beyond the intra-pancreatic bile duct wall, there is no clear distinction between T2 and T3 tumors.¹⁴

Beyond these potential issues with pathologic discrimination between different T-stages within the current classification system, the current study, which represents data from a large, contemporary Western cohort of patients who underwent resection, was not able to differentiate patient survival outcomes based on the current T-stage classification system. This lack of prognostic ability of the current T-stage system has also been reported in previous studies and could in part be attributed to the majority of tumors falling into the T3 category, decreasing the potential discriminative capacity of this system.^{4, 10, 16, 17} In the present study, while extent of tumor invasion (7th AJCC T-stage) was not related to outcomes, tumor size ≥3 cm and presence of LVI were independently associated with decreased survival.

The incorporation of tumor size into AJCC staging is not unprecedented, as size is the primary factor for T-stage of malignancies including breast, lung, and renal cancer, to name a few.¹³ For pancreatic adenocarcinoma, a malignancy in a similar location as distal cholangiocarcinoma, the current T-stage system uses size as a small component of T-stage but is similarly primarily dependent on extent of tumor invasion.¹³ In a recent institutional study of 223 patients who underwent resection of pancreatic adenocarcinoma, a novel T-stage system based only on tumor size better predicted survival than the 7th AJCC system, and this proposed T-stage classification scheme was externally validated in a population from the Surveillance, Epidemiology, and End Results (SEER) database.²⁰ As in pancreatic cancer, the current study showed that in distal cholangiocarcinoma, a tumor size of ≥3 cm was associated with decreased survival. Similarly, DeOliveira et al. reported that in their 30-year single-center experience of cholangiocarcinoma, increased tumor size was independently associated with decreased survival.

In addition to tumor size, presence of LVI was found to be associated with decreased survival in the current study. In other malignancies, LVI has been shown to carry a poor prognosis as well. In patients with colorectal cancer liver metastases, presence of LVI in the primary tumor has been shown to be associated with poor outcomes after liver resection.²¹ Thus, even in the presence of metastatic disease, LVI has been shown to be a marker of poor outcomes. Additionally, in intrahepatic and hilar cholangiocarcinoma, both being tumors histologically related to distal cholangiocarcinoma, presence of LVI has been shown to be independently associated with decreased overall survival even when accounting for other adverse prognostic factors.²²

For distal cholangiocarcinoma, the role of LVI has been debated. In a study of 91 patients from Japan, Kim et al. found that LVI was not associated with survival; however, this study was limited by a small sample size and was of an Eastern population where outcomes may not be directly comparable to those of Western patients. In another study from the East, Kwon et al. reported that for middle and distal bile duct cancers, LVI was associated with decreased overall survival in the context of lymph node negative but not node positive disease.⁴ Others have similarly reported that LVI appears prognostic, but in situations where lymph node disease is present, LVI was no longer associated with survival.¹² In many studies that evaluate tumor-specific factors in distal cholangiocarcinoma, similar to the present one, LVI has been found to be a negative prognostic factor.^{23, 24, 25}

These studies highlight the importance of N-stage in this disease. In creating this novel T-stage, nodal status was intentionally not included in the multivariable model, as the focus was on developing a novel T-stage classification system. As one would expect, each progressive T-stage was more likely to have perineural invasion, lymph node involvement, and a positive resection margin (Table 4B). This highlights the fact that this new proposed T-stage classification system correctly captures the increasing biologic aggressiveness of these tumors which translates to reduced survival after resection. In the assessment of the proposed T-stage system in subset populations of lymph node positive and negative patients, small numbers in the subset populations limited the ability to draw meaningful conclusions with respect to the findings in these subsets, and this will be a focus of future work.

Accurate staging is paramount to effectively care for patients. The proposed T-stage system was thus created by combining tumor size and LVI criteria to better predict patient survival and prognosis for patients with resected disease. A classification system that can predict survival can allow for appropriate counseling of patients and their families. In addition, the ability to anticipate the disease course can help guide the physician in selecting appropriate treatment. Advanced T-stage defined by the proposed system could be used as criteria for administration of adjuvant therapy or potential inclusion in adjuvant therapy trials.

This study was limited by its retrospective design, allowing for assessment of associations but not causality. Additionally, this study was conducted at multiple institutions with no standardized inter-institutional protocols. However, all participating centers were academic hospitals wherein treatment decisions were made in the context of a multi-disciplinary tumor board, and pathologic assessment was conducted by gastrointestinal oncology specialized pathologists. Notably, patients with mid-bile duct tumors treated with bile duct resection alone, technically classified as distal cholangiocarcinoma, were not included. However, these mid-bile duct tumors are rare. This study was designed to assess outcomes of patients who underwent resection where complete pathologic information was available and therefore did not address patients with more advanced disease. This selection permitted assessment of T-stage; however, this small and selected sample size did not allow for additional assessment to create an overall TNM staging system for this disease. This small size reflects the reality of the rarity of this disease in the United States, and external validation of this proposed T-stage system in a distinct patient population is warranted. If validated, incorporation of this new T-stage classification into the next AJCC staging system for distal cholangiocarcinoma could be considered.

Conclusions

For patients with resectable distal cholangiocarcinoma, current AJCC T-stage does not adequately stratify patients with respect to survival. This new proposed T-stage classification system that incorporates tumor size and LVI better differentiates patient outcomes after resection. If validated, incorporation of this new T-stage classification into the next AJCC staging system for distal cholangiocarcinoma could be considered.

Disclosures

None.

Financial support

This study was supported in part by the Katz Foundation.

Conflicts of interest

None declared.

Table 4B.

Pathologic and treatment variables stratified by proposed T-stage classification

Variable	T1 (n = 69)	T2 (n = 82)	T3 (n = 25)	p-Value
Perineural invasion	46 (67.7%)	71 (87.7%)	23 (95.8%)	0.001*
R1 resection margin	9 (13.0%)	15 (18.3%)	12 (48.0%)	0.001*
Lymph node metastasis	24 (34.8%)	55 (67.1%)	19 (76.0%)	<0.001*
High grade	14 (20.9%)	28 (34.6%)	7 (28.0%)	0.185
Adjuvant radiation therapy	21 (35.0%)	44 (55.7%)	12 (54.5%)	0.041*
Adjuvant chemotherapy	34 (56.7%)	57 (71.4%)	16 (72.7%)	0.183

Data reported as n (%); Analysis performed by Chi-squared tests; *p < 0.05.