The roles of the gallbladder and cystic duct (CD) invasions in distal bile duct carcinoma (DBDC) have not been well elucidated.
To define the characteristics and prognostic significance of gallbladder or CD invasions in patients with DBDC.
Organ invasion patterns with clinicopathologic features were assessed in 258 resected DBDCs.
CD invasions (N = 31) were associated with frequent concomitant pancreatic and/or duodenal invasions (23 of 31, 74%) and showed stromal infiltration (16 of 31, 52%) and intraductal cancerization (15 of 31, 48%) patterns. In only 2 cases, invasions with intraductal cancerization were observed in the gallbladder neck. Conversely, all pancreatic (N = 175) and duodenal (83) invasions developed through stromal infiltration. CD invasions were associated with larger tumor size (P = .001), bile duct margin positivity (P = .001), perineural invasions (P = .04), and higher N categories (P = .007). Patients with pancreatic or duodenal invasions had significantly lower survival rates than those without pancreatic (median, 31.0 versus 93.9 months) or duodenal (27.5 versus 56.8 months, P < .001, both) invasions. However, those with gallbladder or CD invasions did not have different survival times (P = .13). Patients with concomitant gallbladder/CD and pancreatic/duodenal invasions demonstrated significantly lower survival rates than those without organ invasions (P < .001).
Gallbladder invasions were rare in DBDCs as neck invasions with intraductal cancerization. CD invasions occurred by stromal infiltrations and intraductal cancerization, whereas all pancreatic and duodenal invasions had stromal infiltration patterns. Gallbladder and/or CD invasions did not affect survival rates of patients with DBDC, while pancreatic and duodenal invasions affected survival rates. Therefore, these differences in survival rates may originate from the different invasive patterns of DBDCs.
Bile duct carcinomas (also known as cholangiocarcinomas) comprise 3% of all gastrointestinal cancers globally.1 Bile duct carcinomas are classified as intrahepatic or extrahepatic bile duct (EBD) carcinomas, and EBD carcinomas are further subclassified as perihilar or distal bile duct carcinomas (DBDCs).2 It was estimated that 10,910 Americans would be diagnosed with EBD or gallbladder carcinomas in 2015.3 Incidences of EBD cancers in Korea are 3.3 and 1.5 per 100 000 men and women, respectively, and are 5 to 7 times higher than those in America.4 About 30% of all bile duct carcinomas are DBDCs.2 Cancer staging of DBDCs was separated from that of proximal or perihilar bile duct carcinomas in the 7th edition of the staging manual of the American Joint Committee on Cancer (AJCC).2,5
In the 7th edition of the AJCC staging system for DBDC, gallbladder invasions were categorized as T3 cancers, similar to pancreatic and duodenal invasions.2 This T3 category was used as an AJCC staging classification in the 3rd to 7th editions.2,6–9 Hong et al10 previously proposed that the depth of the invasion is a more powerful prognostic predictor than the classic AJCC T-category components, such as tumor involvements within or beyond the bile duct wall or pancreas. This proposal was accepted and the measured depths of invasions into the bile duct were included in the T category for DBDCs in the 8th edition of the AJCC staging system.5 Additionally, the 8th edition characterizations for DBDCs recommended that pathologic reports describe the invasion status in adjacent organs including the pancreas, duodenum, gallbladder, colon, stomach, and omentum.5
Few previous studies address the clinical significance of pancreatic or duodenal invasions of DBDCs.11–14 Moreover, to the best of our knowledge, no previous studies describe gallbladder invasions of DBDC or evaluate the associated clinicopathologic and prognostic significance.
Anatomically, the gallbladder can be divided into the fundus, the body, and the neck. The cystic duct (CD) connects the neck of the gallbladder to the common hepatic and bile ducts.2,5 Although the CD is not a part of the gallbladder, it is uncertain whether CD involvement in DBDC is considered a gallbladder invasion because primary carcinomas in the CD have been classified as EBD or gallbladder carcinomas.6–9 In older versions of the AJCC staging manuals, a primary CD carcinoma was categorized as an EBD cancer rather than as gallbladder cancer,6–9 although, from the 7th edition of the AJCC staging system, a primary CD carcinoma follows gallbladder cancer staging.2,5 Therefore, CD invasions of DBDCs may indicate gallbladder involvement.
When DBDCs spread into surrounding organs, such as the pancreas or the duodenum, cancer cells usually directly infiltrate into stromal tissues. In addition to this conventional mode of stromal invasion, intraductal cancerization (also known as intraductal or superficial spreading) has been defined as the growth of cancer cells from stromal tissue into and extending along the preexisting ductal architecture, and has recently been seen in pancreatobiliary tract cancers.15–17 Morphologic distinctions between intraductal cancerization and high-grade intraepithelial neoplasia are, however, almost impossible.15,18 Recently, p53 and/or Smad4 immunostaining analyses have helped to distinguish lesions of the pancreas15,19 because under the conditions of intraductal spreading, cancer cells have a high prevalence of aberrant p53 and/or Smad4 expression and resemble adjacent pancreatic ductal adenocarcinomas (PDCAs), whereas dysplastic cells in high-grade pancreatic intraepithelial neoplasias (PanINs) do not.15,19 Because DBDC and PDCA share a common carcinogenic process and biliary intraepithelial neoplasias (BilINs) are considered as the biliary counterpart of PanINs,20 the above immunostaining analyses could be applied to differentiate between intraductal cancerization of DBDC and high-grade BilINs.
In the present study, we investigated the invasive patterns of gallbladder and CD invasions in detail and evaluated their clinicopathologic and prognostic relevance to DBDCs by comparing pancreatic and duodenal invasions.
MATERIALS AND METHODS
After approval from the Institutional Review Board (2013-0527), a cohort of 258 patients who received surgical resections of primary DBDCs between January 2008 and December 2015 from a single institution was compared in terms of various clinicopathologic variables including patient survival. All cases received pancreaticoduodenectomies with cholecystectomies, including pylorus-preserving pancreaticoduodenectomies or classic Whipple operations with cholecystectomies. All carcinomas originated in the mucosa of bile ducts and from the junction of the CD–common bile duct to the ampulla of Vater. Carcinomas arising in the ampulla of Vater or the pancreas were excluded. Patients with DBDC who were treated with bile duct resections were not included in this study.
Clinical data were obtained for sex, age, operation date, most recent follow-up date, and survival status. Pathologic data included size and specified location of the tumor; growth pattern (papillary, nodular, or diffusely infiltrative patterns); histologic subtype; tumor grade (low [well to moderately differentiated) and high (poorly differentiated and undifferentiated]); the margin status of the bile duct; perineural and lymphovascular invasion; pancreatic, duodenal, gallbladder, and CD invasions; nodal metastasis; and T and N categories and stage grouping according to the 8th AJCC cancer staging system.5 Organ invasion was defined as the direct spread of cancer arising from the distal segment of the common bile duct and into the pancreas, duodenum, or gallbladder.5 When cancer cells were present in the representative sections of CD of either the gallbladder or the distal bile duct, it was considered as a CD invasion. We classified DBDC cases into the following 4 groups on the basis of combined organ invasion patterns: (1) tumors without any invasion of the gallbladder and/or CD, pancreas, or duodenum; (2) tumors with gallbladder and/or CD invasions; (3) tumors with pancreatic and/or duodenal invasions; and (4) tumors with concomitant gallbladder/CD and pancreatic/duodenal invasions. In addition, we categorized organ invasions morphologically into stromal infiltrations and suspicious intraductal cancerizations. Initially, we evaluated the presence of conventional stromal infiltration of cancer cells (Figure 1, A). When we could not identify the foci of stromal invasions, we evaluated the presence of suspicious intraductal cancerization along the biliary tract (Figure 1, B and C).15
To distinguish between intraductal cancerization of DBDC and high-grade BilINs, we immunolabeled p53 and Smad4 for cases of suspicious intraductal cancerization after hematoxylin-eosin staining. Primary antibodies for p53 (1:1500; DO-7, DAKO, Glostrup, Denmark) and DPC4/Smad4 (1:100; EP618Y, Epitomics, Burlingame, California) and a Benchmark autostainer (Ventana Medical Systems, Tucson, Arizona) were used according to the manufacturer's protocols. Immunoreactivity was independently evaluated by 2 pathologists who were blinded to clinicopathologic information.
Statistical analyses were performed with SPSS software (version 17.0, SPSS Inc, Chicago, Illinois). Categorical data were evaluated by the Student t test, the χ2 test, or Fisher exact test. Survival curves were estimated by the Kaplan–Meier method, and the log-rank test was used to calculate associations between survival rates and DBDC. Differences were considered significant when P < .05.
Characteristics of the Cases
The mean age of patients was 65.3 ± 9.3 years, and the male to female ratio was 1:7. The follow-up period after surgical resections ranged from 0.9 to 113.0 months (median, 37.7 months).
Organ Invasions Into Gallbladder and/or CD, Pancreas, and Duodenum
The presence or absence of the pancreatic, duodenal, gallbladder, and/or CD invasions was discernable in all 258 cases. DBDCs involved the pancreas (175 cases, 68%), duodenum (83, 32%), and/or CD (31, 12%) (Figure 2). Among patients with CD invasions, concomitant pancreatic and/or duodenal invasions were identified in 74% (23 of 31) of cases, 15 of which had CD and pancreas involvement and 8 had invasions in the CD, pancreas, and duodenum. All 175 cases of pancreatic invasions and 83 cases of duodenal invasions showed stromal infiltration patterns of invasive DBDCs. In contrast, both stromal infiltration (16 of 31, 52%) and suspected intraductal cancerization (15 of 31, 48%) patterns were noted in CD invasions. Of 23 DBDCs with concomitant CD and pancreatic/duodenal invasions, CDs were more commonly involved via stromal infiltration (13 cases, 57%) than via suspected intraductal cancerization (10, 43%). In contrast, in 8 DBDC cases with CD invasions only, CDs were more commonly involved via suspected intraductal cancerization (5 cases, 63%) than via stromal infiltration (3, 37%). Among 31 cases of DBDCs with CD invasions, only 2 cases with suspected intraductal cancerization invasion patterns were observed in the gallbladder neck; the pancreatic invasion was concomitantly combined in 1 case and no duodenal invasion was recorded. There were no DBDC cases with gallbladder and/or CD invasions showing both stromal infiltration and suspected intraductal cancerization patterns.
Intraductal cancerization was morphologically suspected in 15 DBDC cases with CD invasions, including the 2 cases with gallbladder neck invasions. Cancer cells within the intraepithelial compartments in all CD/gallbladder lesions (intraductal cancerization) were histologically similar to the stromal-infiltrating cancer cells in adjacent DBDCs. To exclude high-grade BilINs from true intraductal cancerizations, p53 and Smad4 immunohistochemical staining analyses were performed both in suspected intraductal cancerization and in stromal infiltration components of all 15 DBDC cases with CD invasions via an intraductal cancerization pattern. All 15 cases had matched aberrant p53 immunolabeling in CD and gallbladder lesions and in invasive DBDCs. P53 was overexpressed in 11 cases (73%; Figure 3, A) and was completely lacking in 4 cases (27%; Figure 3, B). We observed a matched loss of Smad4 expression in both CD/gallbladder lesions and invasive DBDCs in all 15 cases (Figure 3, C). These findings were consistent with ductal cancerization of DBDCs. Therefore, all 15 cases were diagnosed with intraductal cancerization components of DBDCs rather than high-grade BilINs.
Association Between Clinicopathologic Factors and Organ Invasion Status
Because there were only 2 DBDC cases with gallbladder neck invasions and they had combined CD invasions, we focused our investigations on CD invasions. Relationships between clinicopathologic factors and organ invasions, including CD, pancreas, and duodenum involvement in patients with DBDC, are summarized in Table 1. CD invasions were associated with large tumor size (P = .001), involvement of the bile duct resection margin (P = .001), perineural invasion (P = .04), and nodal metastasis (P = .003). CD invasions were not significantly associated with pancreatic and duodenal invasions (P = .42, both). Based on the 8th AJCC staging system, CD invasions were not significantly correlated with the T category (P = .28) but were associated with the N category (P = .007), which was not related to the stage grouping of DBDC (P = .68).
Conversely, pancreatic invasions were closely related to DBDCs with duodenal invasions (P < .001). Pancreatic and duodenal invasions were more frequently observed in tumors showing nodular or diffusely infiltrative growth patterns (P < .001 and P = .02, respectively), lymphovascular invasion (P = .01 and P < .001, respectively), and nodal metastasis (P = .001 and P < .001, respectively). In addition, pancreatic and duodenal invasions were more commonly seen in DBDCs with higher T (P < .001, both) and N categories (P = .002 and P < .001, respectively) as well as higher stage grouping (P < .001, both) of the 8th AJCC staging system. Tumors with pancreatic invasions more frequently exhibited perineural invasions (P < .001). Duodenal invasions were inversely correlated with the involvement of the bile duct margins (P = .03).
Associations Between Clinicopathologic Factors and CD Invasions Based on Invasion Patterns
We evaluated the association between CD invasive patterns and clinicopathologic factors among 31 DBDC cases with CD invasions. However, we could not find any significant difference in clinicopathologic characteristics according to the CD invasive patterns (16 cases of stromal infiltration versus 15 cases of intraductal spread).
Univariate Analysis of Invasions Into Gallbladder and/or CD, Pancreas, or Duodenum
Overall survival rates of patients with DBDC were not affected by the presence of gallbladder and/or CD invasions (P = .13; Figure 4, A). Median survival times for patients without gallbladder and CD invasions, CD invasions only, and gallbladder and CD invasions were 48.2, 26.5, and 17.5 months, respectively. In pairwise comparisons, no significant difference in the survival rate was observed between patients without gallbladder and CD invasions and those with CD invasions only (P = .13). In addition, similar survival rates were observed for patients with only CD invasions and those with combined gallbladder and CD invasions (P = .34). No significant survival rate differences were identified between patients without gallbladder and CD invasions and those with combined gallbladder and CD invasions (P = .16). In contrast, patients with pancreatic invasions (median, 31.0 months) had significantly lower survival rates than those without pancreatic invasions (93.9 months; P < .001; Figure 4, B). Similarly, patients with duodenal invasions (median, 27.5 months) had significantly shorter overall survival rates than those without duodenal invasions (56.8 months; P < .001; Figure 4, C).
Univariate Analysis Based on Combined Patterns of Invasions Into Gallbladder and/or CD, Pancreas, and Duodenum
Based on the presence of invasions into the gallbladder and/or CD, pancreas, and duodenum, (1) 65 (65 of 258, 25%) cases had no invasions in these organs; (2) 8 (3%) had gallbladder and/or CD invasions; (3) 162 (63%) had pancreatic and/or duodenal invasions; and (4) 23 (9%) had concomitant gallbladder/CD and pancreatic/duodenal invasions. Comparisons of these 4 groups revealed significant differences in overall survival rate (P < .001; Figure 4, D), with median survival times of 93.9, 43.1, 32.0, and 25.4 months, respectively. In pairwise comparisons, patients with concomitant invasions and those with pancreatic and/or duodenal invasions had significantly lower survival rates than those without organ invasions (P < .001, both). No significant difference in survival rate was identified between patients with gallbladder and/or CD invasions and those without organ invasions (P = .11). Similarly, no significant difference in survival rate was observed between patients with gallbladder and/or CD invasions and those with pancreatic and/or duodenal invasions (P = .48), or between those with pancreatic and/or duodenal invasions and those with concomitant gallbladder/CD and pancreatic/duodenal invasions (P = .25).
In further analyses, we grouped DBDCs with pancreatic/duodenal invasions and those with concomitant gallbladder/CD and pancreatic/duodenal invasions together to produce the following 3 groups: DBDCs without any organ invasions, DBDCs with gallbladder and/or CD invasions only, and DBDCs with pancreatic and/or duodenal invasions. Significant differences in overall survival rate were identified among these groups (P < .001; Figure 4, E). Moreover, pairwise comparisons revealed that patients with pancreatic and/or duodenal invasions had significantly lower survival rates than those without organ invasions (P < .001). Yet no significant difference in survival rate was found between patients with gallbladder and/or CD invasions only and those with pancreatic and/or duodenal invasions (P = .43), or between those with gallbladder and/or CD invasions only and those without organ invasions (P = .11).
Univariate Analyses of Clinicopathologic Factors
In addition to combined patterns of organ involvement, poor survival rates were associated with growth patterns (P < .001), size (P < .001), and grade (P = .04) of tumors; lymphovascular and perineural invasions (P = .001 and .007, respectively); status of the margins of the bile duct (P = .04); and T and N categories of the 8th AJCC staging system (P < .001, both). However, these univariate analyses revealed no significant associations for survival rates with sex, age, or specified tumor locations.
To identify factors that remained independent predictors, we performed multivariate analyses after adjusting those that were significant in univariate analyses (Table 2). The presence of pancreatic and/or duodenal invasions (P = .03), diffusely infiltrative growth patterns (P = .01), larger tumor size (P = .005), and higher T and N categories (P = .007 and P = .001, respectively) remained independent prognostic predictors of poor overall survival rates for patients with DBDC.
This study is the first to analyze the clinicopathologic significance and survival rate effects of gallbladder and CD invasions in patients with DBDC. Patient survival rates were not significantly affected by the gallbladder and/or CD invasion status, but pancreatic and duodenal invasions were consistently associated with lower patient survival rates. About two-thirds (23 of 31) of DBDC patients with gallbladder and/or CD invasions had concomitant pancreatic and/or duodenal invasions, and these patients had significantly lower survival rates than those without organ invasions. However, overall survival time for patients with gallbladder and/or CD invasions was not significantly different from those with or without concomitant pancreatic/duodenal invasions. Therefore, we conclude that, in contrast to pancreatic or duodenal invasions, gallbladder and/or CD invasions have no prognostic implications for patients with DBDC. In agreement, Ebata et al14 reported that survival rates of DBDC patients with pancreatic or duodenal invasions were lower than for those without pancreatic or duodenal invasions, although they did not include gallbladder invasions in their analyses.
Categorization of gallbladder invasions of EBD cancers varies among editions of the AJCC staging manual. Specifically, gallbladder invasions were categorized as T4 in the 2nd edition,21 but were down-staged to T3 in the 3rd through 6th editions.6–9 From the 7th edition onwards, gallbladder invasions were not considered as belonging to the T category for perihilar bile duct cancers and remained as T3 for DBDCs.1,2 Although gallbladder invasions have traditionally been accepted as a component of the T category–related factors of DBDC, previous studies have not demonstrated the survival effects of gallbladder invasions in DBDC (Table 3).10–14,22–40 Moreover, no single gallbladder invasion was reported in 1543 cases from 23 previous studies of DBDCs. In our institution, gallbladder invasion was extremely rare (2 of 258, 1%) and was associated with CD invasions and intraductal cancerization patterns. Perhaps the low frequency of gallbladder invasion in patients with DBDC reflects the function of the spiral valves of Heister in the CD as an additional anatomical barrier against cancer cell infiltration.
At the point of CD invasion, 48% (15 of 31) of cases were involved via intraductal cancerization of DBDCs. In contrast, all pancreatic and duodenal invasions were stromal infiltrations and were strongly associated with higher T categories; however, stromal infiltrations were more frequently observed among DBDC cases with concomitant pancreas/duodenum and CD invasions. Intraductal cancerization was more often identified among DBDC cases with CD invasions only. We could not identify any significant relationships between CD invasion patterns (stromal infiltration versus intraductal spread) and clinicopathologic characteristics. Finally, we compared overall patient survival rates in DBDC cases with CD invasion patterns of intraductal cancerization and stromal infiltrations but found no significant differences between these groups (data not shown), probably because intraductal cancerization entails growth of cancer cells from stromal tissue into and extending along the biliary tract. To determine whether this observation is significant, further multi-institutional studies with greater numbers of CD invasion cases are required.
In this study, we unexpectedly observed an inverse relationship between duodenal invasions and proximal bile duct margin positivity. A plausible explanation of this inverse relationship may be related to the tumor location of DBDCs (intrapancreatic versus extrapancreatic). Duodenum is closer to the intrapancreatic DBDCs than extrapancreatic DBDCs. In contrast, the proximal bile duct margin is further away from the intrapancreatic DBDCs than extrapancreatic DBDCs. DBDCs of the present study were predominantly composed of intrapancreatic type (245 of 258, 95%), while extrapancreatic DBDCs numbered only 6 (6 of 258, 2%). In intrapancreatic DBDCs, tumors with duodenal invasions (4 of 77, 5%) showed less involvement of the proximal bile duct margin than those without duodenal invasions (36 of 168, 21%; P = .001). In addition, intrapancreatic DBDCs having larger tumor size more commonly showed proximal bile duct positivity (P = .001). However, in extrapancreatic DBDCs, bile duct margin positivity was not significantly related to duodenal invasion and tumor size.
In conclusion, gallbladder invasions of DBDCs were rare and were found as neck invasions with intraductal cancerization. CD invasions were observed in 12% (31 of 258) of DBDCs and followed stromal invasion and intraductal cancerization patterns. In contrast, all pancreatic and duodenal invasions occurred by stromal infiltrations. Pancreatic and duodenal invasions affected survival rates, while gallbladder and/or CD invasions did not. Therefore, the differences in survival rate between patients with gallbladder/CD and pancreatic/duodenal invasions may originate from the different invasive patterns of DBDCs in these organs.
This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2012R1A1A2003360) and by a Grant of Translational R&D Project through Institute for Bio-Medical Convergence, Incheon St. Mary's Hospital, The Catholic University of Korea.
The authors have no relevant financial interest in the products or companies described in this article.
This work was presented in part at the 2016 annual meeting of the United States and Canadian Academy of Pathology; March 12–18, 2016; Seattle, Washington.