Introduction

Papillary tumors of the bile duct are rarely encountered, and have a reported prevalence of 4–15 % among bile duct cancers.1–3 These tumors usually spread superficially,4 and have a better long-term prognosis than other types of bile duct cancer.2,3,5 Meanwhile, multiple papillary tumors termed as “biliary papillomatosis” in a previous version of World Health Organization (WHO) classification6 form a unique disease entity which has been an abiding interest to clinicians because of its frequent recurrence and poor prognosis even after curative resection.7,8 Due to its vague definition in numbers, extent of disease or chronological order, reported malignant potential of biliary papillomatosis ranged from 19.5 to as much as 83 % with conflicting survival outcomes.7–9 Moreover, being proposed as a biliary counterpart of intraductal papillary mucinous neoplasm of the pancreas,10 mucin production observed in papillary bile duct tumors is another concern. However, current version of WHO classification11 completely integrated the term “biliary papilloma(tosis)” and papillary bile duct tumors into “intraductal papillary neoplasm of the bile duct (IPNB),” which excludes consideration for multiplicity and mucin production. Prognosis of IPNB according to its multiplicity or mucin production needs detailed investigation, but the scarcity of IPNB was a great limitation in previous reports.

Therefore, the objectives of this study were to identify the clinicopathologic characteristics of IPNB and to evaluate the prognostic values of multiplicity and mucin production. This study presents 84 consecutive patients with surgically treated IPNB which gives the largest dataset from a single institution.

Methods

This study was approved by our Institutional Review Board, which waived the requirement for informed consent. Clinicopathologic data and radiologic images were prospectively collected in an electronic medical record form, and pathological data were reviewed by a board-certified pathologist (K.B.L.) with 11 years of experience. Between January 2000 and September 2009, 93 patients were pathologically diagnosed as IPNB. Tumors located at the intrahepatic, hilar, and extrahepatic bile ducts were included, and those located at the gallbladder or ampulla of Vater were excluded. After excluding 9 patients who had IPN combined with a mass forming or periductal infiltrative cholangiocarcinoma, a total of 84 patients were finally included for further analysis.

Interpretation of Radiologic Diagnoses

Radiologic IPNB was defined as the presence of intraluminal papillary masses associated with bile duct obstruction or dilatation.12 For radiologic diagnosis, computed tomographic findings were given priority because all patients had CT images. However, magnetic resonance images (n = 69, 82.1 %) and endoscopic or percutaneous direct cholangiograms (n = 73, 87.0 %) were also used for diagnosis. The presence of more than two discontinuous papillary tumors found during any type of radiologic examinations was considered multiple, regardless of tumor size.

Pathologic Assessment of IPNB

IPNB was characterized by dilated bile ducts filled with papillary or villous biliary neoplasm covering delicate fibrovascular stalks.11 Macroscopic tumor morphologies were classified as no definite macroscopic tumor with microscopic IPNB, pedunculated IPNB, sessile IPNB, and as diffuse IPNB in which the entire wall of bile duct was occupied with similar-sized polypoid nodules; hence, no main polypoid mass could be assessed in gross inspection (Fig. 1a–d). Tumors were considered multiple when there were more than two papillary tumors with intervening normal mucosa, or in cases with macroscopically identifiable small nodular lesion around the main mass which does not exceed the full height of main polypoid mass and confines within the mucosa or subepitheial connective tissue in gross inspection (Fig. 1e). Only synchronous multiplicity was taken into consideration in this study because metachronous multiplicity depends upon the duration of follow-up period.

Fig. 1
figure 1

Classification of macroscopic morphologies. a Gross and microscopic findings of microscopic papillary tumors without a definite macroscopic tumor. Intrahepatic bile duct was filled with mucin without any visible mucosal lesion, and microscopic examination revealed stratified columnar epithelium with intracytoplasimic mucin. (H&E staining, ×200). b Gross and microscopic findings of pedunculated tumors (H&E staining, ×12.5). c Gross and microscopic findings of sessile tumors (H&E staining, ×12.5). d Gross and microscopic findings of diffuse papillary tumors (H&E staining, ×12.5). e Definition of the multiplicity of IPNB. Small papillary tumors with a granular surface, separated or combined with the main papillary tumor. These superficial spreading tumors (arrowheads) were considered multiple tumors. Microscopic findings (H&E staining, ×12.5) showed superficial spread of papillary tumors, which correlated well with gross findings of superficial spreading components. Longitudinal section of the common bile duct was made (dashed arrow), and arrows indicate correlating part of gross specimen and microscopic field. f Survival rates for different macroscopic morphologies (n = 84, P = 0.010). According to macroscopic morphology, 5-year survival rates were 100 % for microscopic IPNB, 69.6 % for pedunculated IPNB, 72.7 % for sessile IPNB, and 20.8 % for diffuse IPNB

Full size image

Tumor locations were classified as intrahepatic, perihilar, and distal bile duct13 based on locations identified in surgical specimens. Because T stages differed for intrahepatic and extrahepatic bile duct cancers,14 depth of invasion was classified based on whether tumors were confined to or invaded beyond the bile duct.

Statistical Analysis

IBM SPSS Statistics version 19.0 (IBM Corp., Somers, NY, USA) was used for the analysis. Data fulfilling normal distribution were analyzed as follows: nominal data were compared using the chi-square test, and continuous data using the Student’s t test. Data which were not normally distributed were analyzed with nonparametric tests including Fisher’s exact test or linear-by-linear association for nominal data and Mann–Whitney U test for continuous data. The Kaplan–Meier method was used for survival analysis, and log-rank test was applied. The Cox proportional hazard model analysis was used for multivariate survival analysis. Two-sided P values of less than 0.05 were considered statistically significant, and P values of less than 0.1 were considered marginally significant. Variables with p values of less than 0.05 after univariate analysis were used in the multivariate analysis. Statistical analysis was supervised by Division of Biostatistics, Medical Research Collaborating Center of Seoul National University Hospital Biomedical Research Institute.

Results

Patient Demographics

Patient demographics are listed in Table 1. The study subjects consisted of 63 men and 21 women of mean age 63.6 years at time of diagnosis, and the median duration of follow-up was 41.8 months. Tumors had an intrahepatic (n = 30, 35.7 %), perihilar (n = 24, 28.6 %), or distal bile duct (n = 30, 35.7 %) location. Mucin secretion was identified in 23 patients (28.0 %). Seventy-five patients (89.3 %) had invasive cancer, and 45 patients (53.6 %) had tumors confined to the bile duct. Lymph node metastasis was identified in eight patients (9.5 %). Mean tumor size was 3.5 cm.

Table 1 Patient demographics and tumor characteristics
Full size table

Types of operation included hepatobiliary resection (n = 41, 48.8 %), bile duct resection (n = 18, 21.4 %), and pancreatoduodenectomy (n = 25, 29.8 %). Of the 84 study subjects, 75 (89.3 %) underwent curative resection and 62 of the 75 (82.7 %) achieved R0 resection. Of the nine patients that underwent palliative resection, seven had locally advanced disease and two had diffuse extent of the tumors.

Macroscopic Tumor Characteristics

There were 7 microscopic IPNB, 16 pedunculated IPNB, 49 sessile IPNB, and 12 diffuse IPNB (Table 2). Tumor characteristics were compared in the four macroscopic morphology groups. Microscopic IPNB showed a female preponderance and were located at the intrahepatic bile duct (P = 0.009). Proportion of noninvasive tumors was higher (P < 0.001), and lymph node metastasis was not identified in the microscopic IPNB group. Mucin production was observed in 85.7 % of microscopic IPNB (P = 0.013). Diffuse IPNB were located at perihilar (58.3 %) or at extrahepatic bile ducts (33.3 %), and the lymph node metastasis rate of these tumors (16.7 %) was the highest among four macroscopic morphology groups. The curative resection rate was highest for microscopic IPNB (100 %) and lowest (58.3 %) for diffuse IPNB.

Table 2 Tumor characteristics according macroscopic morphology classification
Full size table

Tumor Characteristics According to Multiplicity

Gross examinations of surgical specimens revealed that 43 patients (51.2 %) had multiple tumors. Comparisons of tumor characteristics with respect to tumor multiplicity are listed in Table 3. No difference in age, sex, tumor location, depth of tumor invasion, lymph node metastasis rate, and mucin secretion was observed between the multiple and single tumor groups. Patients with multiple tumors had a larger total tumor length (P = 0.005) and a lower curative resection rate (P = 0.002). Furthermore, the R0 resection rate was significantly lower in the multiple tumor group (62.8 vs. 87.8 %; P = 0.008).

Table 3 Tumor characteristics according to multiplicity
Full size table

According to radiologic examination findings, 31 patients (36.9 %) had ≥2 separated papillary tumors, and of these, 23 had macroscopically multiple tumors. Therefore, radiologic multiplicity correlated with macroscopic multiplicity in 74.2 %. On the contrary, 23 of the 43 patients with macroscopic multiple tumors were diagnosed to have radiologic multiple tumors; therefore, accuracy of predicting macroscopic multiplicity based on preoperative radiologic imaging findings was 53.5 % with false positive rate of 25.8 % and false negative rage of 37.7 %.

Tumor Characteristics According to Mucin Secretion

Mucin-secreting IPNBs were more frequent in females (43.5 vs. 18.0 %, P = 0.016), and were more commonly located at the intrahepatic bile duct (73.9 vs. 21.3 %, P < 0.001) than the tumors without mucin secretion. Mucin-secreting IPNBs were more confined to the wall of bile duct than invading beyond the bile duct with marginal significance (69.6 vs. 47.5 %, P = 0.071), and lymph node metastasis was less observed in mucin-secreting IPNBs (47.8 vs. 78.7 %, P < 0.001) than the tumors without mucin secretion. Curative resection rate or multiplicity revealed no difference between mucin secreting- or non-secreting IPNBs.

Survival Outcomes of Patients with IPNB

The overall 5-year survival rate for surgically treated IPNB was 64.0 %, with a median survival of 79.7 months. Patients that underwent curative resection had a significantly higher survival rate than those that underwent palliative resection [median survival 105.7 (curative) vs. 45.9 months (palliative), respectively; P = 0.013]. According to macroscopic morphology, 5-year survival rates were 100 % for microscopic IPNB, 69.6 % for pedunculated IPNB, 72.7 % for sessile IPNB, and 20.8 % for diffuse IPNB (P = 0.010, Fig. 1f).

For the 75 patients who underwent curative resection, tumors located at the intrahepatic bile duct had the highest survival rate, followed by tumors located at the distal bile duct and the perihilar bile duct (5-year survival rates, 84.0, 71.3, and 39.3 %, respectively; P = 0.096). Multiple IPNB were found to be associated with significantly lower survival than single IPNB (5-year survival rate 50.7 vs. 85.9 %; P = 0.011, Fig. 2a). Patients with a positive resection margin had lower survival than those with a negative resection margin (5-year survival rates 30.3 vs. 79.9 %; P = 0.010).

Fig. 2
figure 2

a Overall survival rates of IPNB with an associated invasive carcinoma after curative resection according to multiplicity (n = 75, P = .011). The 5-year survival rate of macroscopically multiple tumors was 50.7 % compared with single tumors with 5-year survival rate of 85.9 %. b Disease-free survival rates of IPNB with an associated invasive carcinoma after curative resection according to multiplicity (n = 75, P = .026). The 5-year disease-free survival rate of macroscopically multiple tumors was 36.1 % compared with single tumors with 5-year disease free survival rate of 74.1 %

Full size image

Prognostic Factors of IPNB

Seventy-five patients underwent curative resection for IPNB. Of these patients, univariate and multivariate analyses were performed to identify prognostic factors of IPNBs. As shown in Table 4, a positive resection margin (P = 0.010) and macroscopic multiplicity (P = 0.011) were found to be significantly associated with poorer survival, and depth of tumor invasion (P = 0.146), lymph node metastasis (P = 0.171), macroscopic multiplicity (P = 0.050), and mucin secretion (P = 0.595) were not found to have impact on overall survival of IPNB. Multivariate analysis revealed that a positive resection margin [hazard ratio 2.523, 95 % confidence interval (CI) 1.015–6.274, P = 0.046] and multiplicity (hazard ratio 2.716, 95 % CI 1.055–6.988, P = 0.038) were significant prognostic factors.

Table 4 Univariate and multivariate analysis of prognostic factors of IPNB after curative resection
Full size table

Recurrence Pattern of Invasive Papillary Bile Duct Cancer

After curative resection, 22 patients (29.3 %) experienced recurrence. Recurrence rates were not different in the R0 and R1 resection groups (27.4 vs. 38.5 %, respectively; P = 0.507). Distant metastasis was more frequent (n = 15, 20.0 %) than locoregional (n = 9, 12.0 %) recurrence in all study subjects. Three patients experienced recurrent intraductal papillary cancer and two of the three patients had multiple tumors. Locoregional recurrence was more often observed in patients with multiple tumors than in those without multiple tumors with marginal significance (58.3 vs. 16.7 %; P = 0.089).

After curative resection, the 5-year disease-free survival rate was 58.0 %. The disease-free survival rate was significantly lower in patients with multiple IPNB than in those with single IPNB (5-year disease-free survival rates 36.1 vs. 74.1 %, respectively; P = 0.026, Fig. 2b).

Discussion

Papillary bile duct tumors have not only peculiar gross morphology but also a better long-term prognosis than other types of bile duct cancer.2,3,5 Since 2006, several similarities and differences between intraductal papillary neoplasms of the biliary-pancreatic systems were proposed, and intraductal papillary neoplasms of the bile duct were suggested to be a biliary counterpart of pancreatic intraductal papillary mucinous neoplasm (IPMN).10 Moreover, current version of WHO classification11 unified the term biliary IPMN, biliary papillomtosis, papillary adenocarcinoma of the bile duct, etc. into “intraductal papillary neoplasm of the bile duct (IPNB).”

Although IPNB has characteristic intraductal growth of tumor, the gross morphology of intraductal tumor has a wide variety. In this study, the authors classified the gross morphology of IPNB into four types (microscopic, pedunculated, sessile, and diffuse IPNB), as some radiologists suggested the classification of radiologic morphology of papillary bile duct tumors as polypoid intraductal growth, mucosal spreading growth, cast-like intraductal growth, cystic tumor, and intraductal floating tumors.15 As shown in the data, microscopic IPNB formed distinct clinicopathological subset with excellent prognosis, while diffuse IPNB had worst prognosis among IPNB.

In terms of multiplicity, “biliary papillomatosis” of previous WHO classification6 is now replaced with “intraductal papillary neoplasm” in new WHO classification.11 Clinical importance of multiplicity in IPNB still generates interest among clinicians because of its malignant potential and frequent recurrence. The prognosis of biliary papillomatosis has been reported in a wide variety of malignant potentials which ranged from 19.5 to as much as 83 %.7–9 However, there were confusions concerning its definition.

In this study, we classified macroscopic morphology of IPNB. As shown by our data, a significant portion of IPNB has superficial spreading small papillary tumors (Fig. 1e), and the tumors with presence of these superficial spreading small papillary tumors were designated as multiple IPNB in this study. Previously, these small papillary tumors were noticed by a few studies, which described that the low papillary growth pattern of intraepithelial spread correlated well with macroscopic small granular mucosa,16 or “superficial spread” for tumors was related with macroscopically eroded ductal epithelium.4,17 In this study, 51.2 % of IPNB had multiple tumors and multiple IPNB showed significant correlation with long-term prognosis. Patients with multiple IPNB had lower curative resection rate, and even after curative resection, multiple IPNB had significantly lower overall and disease-free survival compared with single IPNB. Multivariate analysis revealed multiplicity and margin status were independent prognostic factors of IPNB. Moreover, patients with multiple IPNB had more frequent local recurrence with characteristic intraductal recurrence, which constituted 30 % of local recurrence.

The issue of mucin secretion and its clinical impact is another interesting aspect. Recent studies10,18,19 have revealed similarities between IPNB and pancreatic IPMN, and the rate of mucin secretion from cholangiocarcinoma has been reported to range from 13 to 31 %.7,10,19 In the present study, mucin secretion was observed in 28.0 % of IPNB. Moreover, mucin-secreting cholangiocarcinomas have been reported to have higher survival rates than non-mucin-secreting cholangiocarcinomas,19 but this difference resulted from including higher proportion of papillary tumors into mucin-secreting cholangiocarcinomas. As shown in our data, the authors confined the analysis to IPNB, and the overall survival rate of mucin-secreting PBTs did not differ significantly from non-mucin-secreting PBTs, which concurs with a previous large-scale report.7

Although mucin secretion did not impact long-term survival in the present study, patients with a mucin-secreting PBT formed a distinct clinicopathologic subset. This group showed a greater female predominance, and tumors were located in the intrahepatic bile duct in which invasion was confined to the bile duct without lymph node metastasis. These characteristics were demonstrated again in patients with microscopic IPNB. Mucin secretion was identified in 80 % of the patients with microscopic IPNB, and these patients showed female predominance, no patient had invasive cancer, and tumors tended to be located at the intrahepatic bile duct. Furthermore, the survival rate was the highest among four macroscopic morphology groups, and no recurrence was observed among these patients, which concurs with previous reports10,20 which revealed excellent prognoses for noninvasive and minimally invasive papillary tumors. Not only for their pathologic similarity to pancreatic IPMN can microscopic IPNB be designated as a biliary counterpart of pancreatic IPMN because of their clinicopathologic peculiarities and outstanding prognosis.

In summary, IPNB has a favorable prognostic outcome with 5-year survival rate of 64 %. Mucin secretion is identified in 28 %, and the patients with a mucin-secreting IPNB constitute a distinct clinicopathologic subset with a female predominance and less invasive tumor biology, but it does not have any impact on survival outcome. From the standpoints of morphologic morphology, microscopic IPNB has best prognosis followed by pedunculated or sessile IPNB, and diffuse IPNB has the worst prognosis. More than 50 % of IPNB has multiplicity, and patients with multiple IPNB have lower overall and disease-free survival rates and more frequent locoregional recurrence. Positive resection margin and multiplicity were significant independent prognostic factors of IPNB after multivariate analysis. In conclusion, current WHO classification for IPNB needs consideration for macroscopic morphology and multiplicity considering its prognostic impact of IPNB.