Introduction

Intraductal papillary neoplasm of the bile duct (IPNB) was defined by the World Health Organization (WHO) in 2010. The concept of IPNB continues to evolve, and in the latest WHO classification (fifth edition, 2019), it is defined as a grossly visible premalignant neoplasm with intraductal papillary or villous growth of biliary-type epithelium [1]. The prevalence of IPNB is rare, accounting for approximately 4–15 percent of biliary tract tumors, and its pathophysiological characteristics remains incompletely understood [2]. IPNB exhibits unique features distinct from other biliary tract tumors, including papillary growth [3] and mucin production [4], which resemble to those of pancreatic intraductal papillary mucinous neoplasm (IPMN), compared rather than a tumor in biliary tract. And thus, historically, IPNB has been discussed in contrast to pancreatic IPMN due to these similarities [5,6]. In pancreatic IPMN, morphological differences such as branched duct, main duct, and mixed types reflect the degree of malignancy, and based on those morphologies, therapeutic strategies are determined [7]. This is a unique feature among digestive tract tumors. Currently, common criteria for guiding treatment strategies in IPNB have not been established. However, considering IPNB as the counterpart of pancreatic IPMN, on treatment strategy for IPNB, it may be appropriate to refer to morphological classification rather than the TNM classification for biliary tract cancer. Based on this concept, we classified the morphology of IPNB in 2013, reporting their respective characteristics [8]. Several subsequent studies attempting morphological classification like ours have been reported [9–12], but none have yielded including ours results that could guide treatment strategies until now. This may be due to insufficient understanding of tumor characteristics of IPNB, particularly unique features of IPNB that are distinct from pancreatic IPMN such as the complexity of biliary tract anatomy and high malignancy [12].

To further elucidate the clinicopathological characteristics of IPNB and establish a global consensus, a collaborative study on IPNB has been conducted since 2016 by the Japan Biliary Association (JBA) and the Korean Association of Hepato-Biliary-Pancreatic Surgery (KAHBPS) [13]. The newly proposed pathological diagnostic criteria (Type 1 and Type 2 subtypes), based on observations by specialized pathologists, have had the greatest impact on the diagnosis of IPNB in recent years and were adopted in the latest WHO classification of digestive system tumors in 2019 [1].

This study aims to update the morphological classification of IPNB previously reported by us, incorporating the new consensus represented by the Type 1 and Type 2 pathological classifications, and validate its clinical significance in establishing a surgical strategy for IPNB.

Methods

Proposal for a morphologic classification of IPNB: Branched, main duct, and mixed type

IPNB patients were classified into three types—branched, main duct, and mixed—according to our newly proposed morphological criteria, which are based on a combination of cystic and bile duct imaging features (Fig 1A). Fig 1B illustrates each type.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Criteria for the morphologic classification of IPNB.

The IPNBs were morphologically classified into three types: branched, main duct, and mixed types, based on the specific criteria. (A) This criteria combine two independent factors: cystic and bile duct characteristics, which are assessed through preoperative imaging studies. (B) The representative schemas for each morphologic type. Abbreviations: IPNB, intraductal papillary neoplasm of the bile duct.

https://doi.org/10.1371/journal.pone.0325081.g001

Given the close embryological relationship between the biliary tract and the pancreas [17], IPNB can be morphologically classified in a manner analogous to pancreatic IPMN, into branched, main duct, and mixed types [18]. Based on the international criteria for pancreatic IPMN [19], we first defined IPNBs with cystic lesions as the branched type, those with non-obstructive ductal dilatation as the main duct type, and those exhibiting both features as the mixed type. Based on these representative cases, cystic lesions were considered the “cystic factor,” a defining feature of the branched type, while bile duct dilatation without tumor obstruction was considered the “bile duct factor,” a defining feature of the main duct type. Bile duct wall thickening—sometimes accompanied by invasion into surrounding organs and mimicking cholangiocarcinoma [11]—was considered the feature to originate from large bile ducts, and thus, categorized as part of the bile duct factor. Classification criteria for the three morphological subtypes were then established based on the combination of these two factors. Representative clinical cases for each subtype are presented in S3, S4, and S5 Figs.

Results

Clinical and pathological characteristics among three morphologic subtypes of IPNB: Branched, main duct, and mixed

Among all 42 patients with IPNB, 10 (23%) were classified as the branched type, 26 (62%) as the main duct type, and 6 (15%) as the mixed type (Table 1). Half of the patients with the mixed type experienced symptoms such as cholangitis, jaundice, or abdominal pain. Tumor lesion extent significantly differed among the three morphologic types (p < 0.001 for branched vs. main duct; p = 0.012 for branched vs. mixed). All lesions in the branched type were confined to the liver. In contrast, 69% of main duct type cases had extrahepatic tumors, and only 4% showed intrahepatic involvement. Among the mixed type, 17% had tumors in the extrahepatic bile duct, 33% in the intrahepatic duct, and 50% had lesions extending to both regions. Mural nodules were observed in 83% of all IPNB cases (n = 35), with detection rates of 90%, 77%, and 100% in the branched, main duct, and mixed types, respectively. All patients with the branched type underwent hepatectomy, with one case requiring additional extrahepatic bile duct resection. Similarly, all patients with the mixed type underwent hepatectomy, and half of them also required extrahepatic bile duct resection. In the main duct type group, 35% underwent hepatectomy, 62% underwent pancreaticoduodenectomy, and 4% underwent extrahepatic bile duct resection alone. Regarding pathological findings, low-grade intraepithelial neoplasia was more frequently observed in the branched type, whereas invasive carcinoma was more prevalent in the main duct type. Among IPNBs with associated invasive carcinoma, tumor invasion beyond the bile duct wall was observed in 0% of branched, 18% of main duct, and 25% of mixed types. In this population, the rate of residual tumor was significantly higher in the mixed type than in the branched type (75% vs. 0%, p = 0.048). Lymph node metastasis was identified in four patients (18%), all of whom had main duct type IPNB.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patient demographics according to morphologic classification.

https://doi.org/10.1371/journal.pone.0325081.t001

Postoperative survival by morphology and lesion extent in all IPNB cases

In postoperative DSS in patients with the branched type IPNB, there were no postoperative deaths. The 1-, 3-, and 5-year DSS rates for patients with the main duct type were 96%, 90%, and 90%, respectively. No significant difference was observed in DSS between the branched and main duct types. However, the mixed type exhibited significantly lower 1-, 3-, and 5-year DSS rates of 100%, 50%, and 50%, respectively, compared to the branched type (p = 0.044, Fig 2A). As shown in Fig 2B, DSS did not show significant differences among the three groups based on tumor lesion extent (p = 0.167 between intrahepatic and extrahepatic bile duct lesions, p = 0.222 between intrahepatic and both lesions).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Postoperative survival by morphology and lesion extent in all IPNB cases.

DSS in patients with IPNB was compared according to (A) morphological subtype—branched, main duct, and mixed—and (B) anatomical extent of the tumor—confined to the intrahepatic bile duct, extrahepatic bile duct, or both. No disease-specific deaths occurred in the branched type. There was no significant difference in DSS between the branched and main duct types. In contrast, the mixed type showed significantly worse 1-, 3-, and 5-year DSS compared to the branched type (p = 0.044). DSS did not significantly differ among the groups based on tumor extent. Abbreviations: DSS, disease-specific survival; IPNB, intraductal papillary neoplasm of the bile duct.

https://doi.org/10.1371/journal.pone.0325081.g002

Postoperative survival by pathological tumor progression factors in IPNB with associated invasive carcinoma

In thirty-one patients with IPNB with associated invasive carcinoma, DSS was compared based on pathological tumor progression factors. All patients with tumors confined to the mucosal (M) or fibromuscular (FM) layers survived without disease-specific death for 5 years after surgery. In contrast, patients with tumor invasion beyond the bile duct wall (serosal exposure [SE] or serosal invasion [SI]) had 1-, 3-, and 5-year DSS rates of 83%, 0%, and 0%, respectively, which were significantly lower (p < 0.001; Fig 3A). Patients with tumor extension to the subserosal (SS) or serosal (S) layers also had lower DSS compared to those with M or FM invasion (p = 0.037), although the difference became more evident after 5 years postoperatively. In patients with lymph node metastasis, the 1-, 3-, and 5-year DSS were significantly lower compared to those without lymph node metastasis. The DSS rates were 96%, 91%, and 91% versus 100%, 67%, and 67%, respectively (p = 0.021, Fig 3B). In patients with residual tumor (R1 or R2), the 1-, 3-, and 5-year DSS were 86%, 64%, and 64%, respectively. Although these rates were lower than those in patients who underwent curative resection, the difference was not statistically significant (p = 0.119, Fig 3C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Postoperative survival by pathological tumor progression factors in IPNB with associated invasive carcinoma.

In patients with invasive IPNB, DSS was analyzed according to pathological indicators of tumor progression: (A) invasion depth, (B) lymph node metastasis, and (C) residual tumor status. Patients with tumor invasion beyond the bile duct wall (SE or SI) had a 5-year DSS rate of 0%, significantly lower than those with M or FM invasion (p < 0.001). The presence of lymph node metastasis was also associated with significantly worse 5-year DSS (p = 0.021). Patients with residual tumor (R1 or R2) had lower 5-year DSS than those with curative (R0) resection, although this difference did not reach statistical significance (p = 0.119). Abbreviations: DSS, disease-specific survival; IPNB, intraductal papillary neoplasm of the bile duct Tumor depth: M (mucosal)/ FM (fibromuscular)/ SS (subserosa)/ SE (serosal exposure)/ SI (serosal invasion).

https://doi.org/10.1371/journal.pone.0325081.g003

Postoperative recurrence by morphology in IPNB with associated invasive carcinoma

The association between postoperative recurrence and morphological subtypes was evaluated in patients with invasive IPNB. No recurrence was observed in the branched type. In contrast, the mixed type showed a high recurrence incidence, with 1-, 3-, and 5-year cumulative recurrence rates of 25%, 25%, and 62%, respectively (p = 0.074 vs. branched type, Fig 4). In the main duct type, the recurrence risk was intermediate, with 1-, 3-, and 5-year cumulative recurrence rates of 6%, 6%, and 13%, respectively, and no significant difference compared to the branched type (p = 0.246).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Postoperative recurrence by morphology in IPNB with associated invasive carcinoma.

The association between postoperative recurrence and morphological subtypes was evaluated in patients with invasive IPNB. No recurrence was observed in the branched type. In contrast, the mixed type showed a high 5-year recurrence rate of 62%. The main duct type had an intermediate risk, with a 5-year cumulative recurrence rate of 13%. Abbreviations: IPNB, intraductal papillary neoplasm of the bile duct.

https://doi.org/10.1371/journal.pone.0325081.g004

Discussion

The clinical significance of our proposed morphological classification for IPNB lies in three key aspects: (i) preoperative diagnostic capability, (ii) prognostic relevance based on morphological differences, and (iii) simplicity and practicality of the criteria. Given the need for straightforward application in clinical settings, our classification—based solely on cystic and bile duct factors—is highly practical (Fig 1). Notably, it is the first to demonstrate a relationship with pathological classification.

Previous morphological classifications of IPNB considered the lesion location (intrahepatic vs. extrahepatic) as a key factor [8,10,12]. In contrast, our new morphological criteria do not take lesion location into account. This is based on the understanding that IPNB can arise not only from biliary epithelial cells but also from peribiliary glands (PBGs) [20]—accessory glands distributed along both intrahepatic and extrahepatic bile ducts, which contribute to secretion and epithelial repair [18,21]. Lesions derived from biliary epithelium, when arising in small bile ducts (after the segmental branches), typically may present as the branched type and be in the intrahepatic bile ducts. In contrast, PBG-derived lesions may occur irrespective of intrahepatic or extrahepatic bile duct, as PBGs are present throughout the biliary tract. This location-independent approach allowed our classification to include atypical cases that present solely with mural nodules or wall thickening in the bile duct, without cystic changes or non-obstructive ductal dilatation. Previously, such lesions were classified differently based on their anatomical location rather than their morphological features. We believe that these lesions are fundamentally similar in morphology. Therefore, by classifying them into the main duct type based solely on the presence of the bile duct factor, regardless of lesion location, our new criteria resolve this inconsistency (S4 FigB).

Since IPNB is a precancerous lesion that can progress to invasive cancer [22], early detection and surgical resection are essential. However, a definitive surgical strategy has yet to be established. Drawing inspiration from the association between morphology and prognosis in pancreatic IPMN [19], we investigated whether our morphological classification could similarly predict the prognosis of IPNB. Among previous studies that attempted to classify IPNB morphology [9–12,23,24], ours is the first to successfully demonstrate a correlation between morphological subtype and postoperative prognosis. While Luvira et al. [9] stratified prognosis, their study excluded IPNBs in the extrahepatic bile duct—a major limitation. Among our classification—branched, main duct, and mixed—the mixed type had the poorest survival outcomes (Fig 2A). One possible explanation is the higher risk of residual tumor in the mixed type. Due to the extensive spread of lesions—half of which involved both intrahepatic and extrahepatic bile ducts—achieving R0 resection is often difficult (residual tumor rate significantly higher than in the branched type, p = 0.048; Table 1). However, lesion extent alone does not appear to fully account for the poor prognosis associated with the mixed type. As shown in Fig 2B, there were no significant differences in DSS among patients with intrahepatic, extrahepatic, or both intra- and extrahepatic tumor involvement. This suggests that our classification reflects not only the anatomical extent of the tumor but also underlying biological characteristics—for example, its association with pathological subtypes. We next examined the relationship between prognosis and cancer staging. In invasive IPNB, tumor stage is determined according to different classification systems based on tumor location—namely, intrahepatic, perihilar, or distal bile duct cancer—making direct comparison across cases impractical. Instead, we focused on two common oncologic factors: depth of invasion and lymph node metastasis, which can serve as surrogates for pathological staging. As shown in Fig 3, both were strongly associated with poor prognosis and, in terms of p-values, were even more predictive than morphological subtype. However, as these factors are only assessable postoperatively, our morphological classification offers a major advantage by enabling preoperative prognostic assessment and supporting treatment strategy.

We propose several practical implications of our morphological classification for surgeons. While the choice of surgical procedure (e.g., hepatic resection or pancreaticoduodenectomy) primarily depends on the anatomical location and extent of the tumor, our classification can aid in determining the appropriate resection extent. It serves as a rationale for surgical aggressiveness when considering highly invasive procedures such as hepatopancreatoduodenectomy. For example, mixed-type IPNB, associated with poor prognosis, may require wider surgical margins. In contrast, branched-type IPNB, which typically exhibits limited invasion and lower malignancy, may allow for more limited resection and is often amenable to minimally invasive approaches such as robotic or laparoscopic surgery. Postoperatively, the classification may also inform the need for adjuvant therapy and surveillance. The mixed type demonstrated a significantly higher recurrence rate, with over half of the patients experiencing recurrence within five years (Fig 4). In contrast, no recurrence was observed in the branched type, even when invasive components were present. These findings suggest that mixed-type IPNB may benefit from adjuvant therapy and intensive follow-up, while such measures might be unnecessary in the branched type. Crucially, this classification provides prognostic information preoperatively, supporting personalized treatment planning from the outset. Surveillance strategies like pancreatic IPMN are not recommended, even for the branched type. We encountered a case initially classified as branched type that progressed over 14 years to mixed type with diffuse bile duct involvement (S7 and S8 Figs). Given the difficulty of curative resection in advanced mixed-type IPNB, timely surgery is advisable.

To objectively evaluate the still poorly understood nature of IPNB, the pathological subtypes, Type 1 and Type 2, were proposed [13]. Some differences between these subtypes have been reported, with Type 2 showing higher malignancy (over 80% of Type 2 exhibiting invasive carcinoma compared to 30% in Type 1, p = 0.007, S1 Table). Many previous studies have reported worse postoperative survival in Type 2 IPNB compared to Type 1 [25–28]. Although our results did not show statistical significance, the disease-specific survival in Type 2 was lower than in Type 1, with a 5-year survival rate of 100% for Type 1 versus 76% for Type 2 (S6 Fig). While this pathological classification should be more critical in future clinical settings as a basis, it has limitations for preoperative management since the types can only be determined postoperatively. Our novel morphological classification is, to the best of our knowledge, the first tool to demonstrate a correlation between morphological and pathological features. In our cohort, Type 1 exhibited more branched type morphology than Type 2 (60% vs. 18%), while Type 2 exhibited more main duct type morphology than Type 1 (64% vs. 20%) (p = 0.039, S1 Table). Our morphological classification may complement the pathological classification by predicting its subtype preoperatively.

Limitations of this study include its retrospective, single-center design; however, the most critical limitation is the small cohort size. While we demonstrated an association between our morphological classification and patient prognosis, the small number of cases—particularly the limited number of mixed-type IPNBs with poor prognosis (n = 6)—precludes definitive conclusions about its generalizability. For instance, to establish the classification as an independent prognostic factor, it would be necessary to compare it with known pathological risk factors such as depth of invasion and lymph node metastasis using multivariate analysis, which is not feasible given the current sample size. Nonetheless, we believe that the concept and criteria of our novel morphological classification are of significant value and hold considerable potential in the clinical setting. Future multicenter studies with larger cohorts are needed to objectively validate the significance and reliability of this classification system.

In conclusion, we classified IPNB into three morphological types—branched, main duct, and mixed—drawing inspiration from the classification of pancreatic IPMN. This classification demonstrated a clear correlation with postoperative prognosis and may help guide surgical strategies tailored to each morphological type in patients with IPNB.

Supporting information

Acknowledgments

Not applicable.