Prognostic Significance of Intratumoral and Peritumoral Budding in Distal Extrahepatic Bile Duct Carcinoma Open Access

Distal extrahepatic bile duct carcinoma (DBDC) is an aggressive neoplasm with a poor prognosis [1]. DBDC staging was separated from that of proximal extrahepatic bile duct carcinoma (PBDC) in the 7th edition of the American Joint Committee on Cancer (AJCC) staging manual [2]. Furthermore, in the 8th edition, the DBDC T category of the staging system was modified according to tumor invasion depth as T1 for <5 mm, T2 for 5–12 mm, and T3 for >12 mm, and the N category was classified into three tiers according to the number of metastatic nodes as N0 for no lymph node involvement, N1 for 1−3 metastatic nodes, and N2 for ≥4 metastatic nodes [3].

Tumor budding (TB) is histologically defined as single cancer cells or clusters ≤4 cancer cells and biologically represents the initial phase of alterations for tumor invasion through epithelial-mesenchymal transition [4]. The assessment of TB was standardized by the hotspot method of the International Tumor Budding Consensus Conference (ITBCC) in 2016 [5]. TB has been most prominent at the invasive front of tumors, which has generally been termed “peritumoral budding (PTB)” [5]. Conversely, TB can also be found within the main tumor body, which is referred to as “intratumoral budding (ITB)” [6]. PTB can only be evaluated at the invasive front of tumors in surgical specimens, whereas ITB can be assessed in both biopsy and surgically resected specimens [7, 8]. PTB has been recognized as a representative adverse prognosticator in colorectal carcinoma (CRC) [4, 5]. PTB was associated with higher stage and tumor grade, lymphovascular invasion, and nodal and distant metastases [5]. Specifically, PTB was predictive for nodal metastasis in patients with pT1 CRC and, thereby, aided in selecting radical surgery with node dissection [4, 9]. PTB indicates a high-risk group in patients with stage II CRC, for which adjuvant chemotherapy should be considered [4, 10]. Emerging evidence also suggests that ITB detected in the pretreatment biopsies of CRC patients could predict nodal and distant metastases and poor responses to neoadjuvant therapy [4, 7, 8].

TB, specifically PTB, has been reported as a prognostic predictor in patients with bile duct carcinomas (BDCs) [11‒17]. However, only two studies of TB in DBDCs applied the ITBCC criteria and the 8th edition DBDC staging scheme to a small number of cases (Table 1) [11, 12], and the prognostic utility of ITB has not been elucidated in DBDC. This study investigated the associations of TB, both PTB and ITB, with clinicopathologic factors and analyzed the prognostic implications of TB in patients with DBDC.

A total of 410 surgically resected primary DBDCs collected from a previous study were included from two institutions (Asan Medical Center between 2008 and 2015 and Incheon St. Mary’s Hospital between 2001 and 2013) [18]. Cases with neoadjuvant chemotherapy were excluded as previously described [18]. Clinicopathologic DBDC data obtained from a previous study were used, including T and N categories and staging group based on the 8th edition of the AJCC scheme [18]. In addition, the overall survival (OS) time and the survival status of the patients were updated [18].

PTB and ITB were independently counted in one hotspot of a ×20 objective field (area, 0.785 mm²) at the peritumoral invasion front and in the intratumoral area, respectively. PTB and ITB were graded based on the number of TBs as G1 (0–4), G2 (5–9), or G3 (≥10), as recommended by the ITBCC (Fig. 1a, b) [5]. Consequently, high PTB and ITB grades (PTB^High and ITB^High) were determined to maximize their sensitivities and specificities in predicting OS using receiver operating characteristic (ROC) curve analysis as described in previous studies [19, 20].

SPSS Statistics for Windows (version 28.0; IBM, Armonk, NY, USA) was used for the analyses. Continuous and categorical variables were compared using the Student’s t test and χ² and/or Fisher’s exact test. Survival curves were plotted using the Kaplan-Meier method, and statistical significances were tested using the log-rank test and the Cox proportional hazards model. A p value of <0.05 was considered statistically significant.

Table 2 shows the clinicopathologic characteristics of the patients. The median follow-up time of the patients after surgery was 36.7 ± 25.2 months (range: 1.1−119.6 months). Histologically, DBDCs were categorized into 380 tubular adenocarcinomas (92.7%), one mucinous carcinoma (0.2%), 23 adenocarcinomas arising from the intraductal papillary neoplasm of the bile duct (IPNB) (5.6%), two adenosquamous carcinomas (0.5%), and four undifferentiated carcinomas (1.0%). The T categories included 4 Tis (1.0%), 125 T1 (30.5%), 201 T2 (49.0%), and 80 T3 (19.5%). No T4 category was seen in our cohort. Nodal metastases were seen in 145 cases (35.4%), with 112 N1 (27.3%) and 33 N2 (8.1%) cases. Finally, there were four cases of stage 0 (1.0%), 106 of stage I (25.9%), 142 of stage IIA (34.6%), 124 of stage IIB (30.2%), 32 of stage IIIA (7.8%), and two of stage IV (0.5%), with no cases of stage IIIB.

The incidence of PTB was varied, with average numbers of 7.9 ± 4.1. The PTB grades consisted of G1 (94 cases, 22.9%), G2 (59, 14.4%), and G3 (257, 62.7%). After ROC analysis, PTB^High was defined as PTB G2 and G3. The average number of ITB cases was 7.3 ± 4.5, and the ITB grades included G1 (123 cases, 30.0%), G2 (49, 12.0%), and G3 (238, 58.0%). ITB^High was defined as ITB G3 by ROC analysis. Consequently, PTB^High and ITB^High were identified in 316 (77.1%) and 238 (58.0%) cases, respectively.

Table 3 shows the association of PTB and ITB with clinicopathologic factors. PTB^High was significantly associated with DBDCs unrelated to IPNB (p < 0.001), a sclerosing growth pattern and higher tumor grade (both p < 0.001), lymphovascular invasion (p = 0.011), perineural and pancreatic invasion (both p < 0.001), nodal metastasis (p = 0.001), higher T and N categories (p < 0.001 and p = 0.002, respectively), and staging group (p < 0.001). Similar to PTB^High, ITB^High was more commonly found in DBDCs unrelated to IPNB (p < 0.001), tumors showing a sclerosing growth pattern and higher tumor grade (both p < 0.001), lymphovascular invasion (p = 0.022), perineural and pancreatic invasion (both p < 0.001), and higher T category and staging group (both p < 0.001). In addition, ITB^High was correlated with duodenal invasion (p = 0.017).

Table 4 shows the relationship between clinicopathologic factors and OS. DBDC patients with PTB^High and ITB^High had significantly shorter OS times than those with PTB^Low and ITB^Low (both p = 0.001; Fig. 2a, b). In addition, shorter OS was associated with DBDCs showing a larger size and sclerosing growth pattern (both p < 0.001), higher histologic grade (p = 0.006), extrapancreatic location (p = 0.015), adenocarcinomas unrelated to IPNB (p = 0.008), pancreatic and duodenal invasion (both p < 0.001), lymphovascular (p < 0.001) and perineural invasion (p = 0.003), cancer involvement of the bile duct resection margin (p < 0.001), nodal metastasis (p < 0.001), and higher T and N categories and disease stages (all p < 0.001). Patient age and sex did not affect OS.

We further evaluated the independent prognostic impact of PTB^High and ITB^High and other clinicopathologic factors considered significant by univariate analyses (Table 5). Since both PTB^High and ITB^High were types of TB, PTB^High or ITB^High was alternatively included and compared to other clinicopathologic factors. According to the multivariate analysis of PTB and clinicopathologic factors, PTB^High (p = 0.040), larger tumor size (p < 0.001), sclerosing growth pattern (p = 0.011), duodenal invasion (p = 0.019), extrapancreatic tumor location (p = 0.002), involvement of the bile duct margin (p = 0.001), and higher staging group (p < 0.001) were poor independent prognostic predictors of OS in DBDC patients. When the prognostic significance of ITB^High and clinicopathologic factors were independently analyzed, ITB^High (p = 0.022), as well as larger tumor size (p = 0.002), sclerosing growth pattern (p = 0.016), extrapancreatic tumor location (p = 0.004), involvement of the bile duct margin (p = 0.005), and higher T and N categories (p = 0.021 and p < 0.001, respectively), remained poor independent prognostic indicators of OS in DBDC patients.

When the prognostic significance of PTB and ITB was analyzed depending on the T category (Fig. 3a–f), significant survival differences were observed only for ITB. In the T1 category (n = 125), the survival time of patients with ITB^High (median, 59.9 months) was significantly shorter than those with ITB^Low (107.0 months; p = 0.013). However, in the T2 (n = 201) and T3 categories (n = 80), there was no significant difference in survival time between patients with ITB^High and ITB^Low (37.9 vs. 41.9 months, p = 0.278 in T2; 20.7 vs. 24.6 months, p = 0.410 in T3). In contrast, no relationship between PTB and OS was seen in the T1 (61.3 vs. 107.0 months; p = 0.105) and T2 groups (38.1 vs. 56.8 months; p = 0.429). In the T3 group, there was a borderline significant difference in the survival time between patients with PTB^High (21.2 months) and PTB^Low (39.8 months; p = 0.054).

We also compared patient survival based on the N category (Fig. 4a–f). In DBDCs with N0 category (n = 265), patients with ITB^High (median, 50.3 months) had significantly shorter survival times than those with ITB^Low (107.0 months; p = 0.001). However, ITB was not related to survival in patients with tumors in the N1 (n = 112) and N2 categories (n = 33) (20.8 months in ITB^High and 33.1 months in ITB^Low, p = 0.460 in N1) (17.9 months in ITB^High and 15.7 months in ITB^Low, p = 0.318 in N2). Similarly, patients with DBDCs in the N0 category with PTB^High (55.7 months) had shorter survival times than those with PTB^Low (79.6 months; p = 0.041). In contrast, no significant difference was seen in the survival time between patients with PTB^High and PTB^Low in the N1 (22.6 months vs. 33.1 months; p = 0.479) and N2 categories (17.0 months vs. 11.5 months; p = 0.931).

In terms of disease stages, significant survival differences were observed only in ITB (Fig. 5a–d). In stage I (n = 106), patients with ITB^High had shorter survival times (median, 61.3 months) than those with ITB^Low (107.0 months; p = 0.025). ITB was not associated with patient survival in each disease stage, II (n = 266), III (n = 32), and IV (n = 2) (p = 0.189, 0.295, and 0.317, respectively). For PTB (Fig. 6a–d), the survival time of patients with stage I DBDC and PTB^High tended to be shorter than those with PTB^Low, but it was not statistically significant (107.0 vs. 78.8 months; p = 0.297). Disease stages II, III, and IV in patients with PTB were not associated with survival (p = 0.327, 0.945, and 0.317, respectively). All DBDCs in stage I were T1N0.

We analyzed the survival data based on the combination of PTB and ITB levels (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000535847). When categorized into 4 subgroups of PTB^Low/ITB^Low, PTB^High/ITB^Low, PTB^Low/ITB^High, and PTB^High/ITB^High, the combined variable significantly stratified patient survival (p = 0.002). Patients with PTB^Low/ITB^Low (90 cases, 22.0%) had a median OS time of 75.6 months, whereas those with PTB^High/ITB^High (234, 57.0%) had shorter median OS time of 35.1 months with an almost 2-fold increased risk of death. Patients with PTB^High/ITB^Low (82, 20.0%) and PTB^Low/ITB^High (4, 1.0%) had median survival times of 48.2 and 29.9 months, respectively.

In further analysis, we combined PTB^High/ITB^Low and PTB^Low/ITB^High together to categorize cases into 3 groups: PTB^Low/ITB^Low (90 cases, 22.0%), either PTB^High or ITB^High (86, 21.0%), and PTB^High/ITB^High (234, 57.0%). Patient survival rates were significantly associated with the 3 PTB/ITB groups with the median OS times of 75.6, 41.9, and 35.1 months in PTB^Low/ITB^Low, either PTB^High or ITB^High, and PTB^High/ITB^High, respectively (p = 0.001). When the prognostic significance of these 3 PTB/ITB groups was analyzed based on the disease stage, there were no significant differences in survival (p = 0.116 in stage I; p = 0.431 in stage II; p = 0.419 in stage III; p = 0.317 in stage IV).

The present study investigated ITB and PTB in a large number of patients (n = 410) and found strong prognostic predictability. Most previous studies of PTB in BDCs included intrahepatic BDCs and/or PBDCs without the specific discrimination of DBDC groups (Table 1) [11‒17]. All analyses were performed in hematoxylin and eosin-stained whole tissue specimens with applied cut-offs for PTB^High in G2 and more on the basis of ITBCC 2016 [11‒17], which was identical to our study. Among those studies, only Nakayama et al. [11] and Ogino et al. [12] analyzed the prognostic significance of PTB^High in DBDC cases (<120 cases) according to the 8th TNM classification. The overall prevalence of PTB^High in these two DBDC studies was 81% (145/180), all of which had significant correlations with the OS. Similarly, we found PTB^High in 77% (316/410) in our DBDC cohort and revealed its predictability for OS. Ogino et al. [12] demonstrated an association between PTB^High and aggressive pathologic features, such as advanced T category, poor histologic differentiation, lymphovascular and perineural invasion, and nodal metastasis, but they did not describe its association with the staging group [12]. Consistent with our present study, Nakayama and colleagues found that PTB^High was correlated with higher T and N categories and the staging group, higher tumor grade, lymphovascular and perineural invasion, and nodal metastasis [11]. In the present study, we newly found that PTB^High was more commonly present in DBDCs unrelated to IPNB and tumors showing sclerosing growth patterns and pancreatic invasion. More importantly, we found that ITB^High, like PTB^High, was also strongly associated with aggressive pathologic factors and poor prognosis in DBDC. Therefore, we proposed ITB as a surrogate prognostic indicator for DBDC patients, comparable to PTB. Interestingly, ITB^High correlated with the duodenal invasion of DBDC.

No standardized methodologies or cut-offs for defining high-grade ITB existed. Nevertheless, accumulating evidence indicated that the assessment of ITB could have prognostic utility across several different tumor types, such as carcinomas of the colorectum, tongue and oral floor, and head and neck [4, 7, 8, 21–23]. In this present study, we constructed a cut-off value for ITB to maximize the sensitivity and specificity in predicting the OS, and ITB^High was finally defined as G3. The sensitivity was 65% and the specificity was 51% at an ITB cut-off of G3. Although ITB had poor accuracy in discriminating patient survival with an area under the curve (AUC) of only 0.58 in DBDCs, ITB^High significantly predicted aggressive tumor behavior and worse OS. When using an ITB cut-off of G2, the sensitivity increased to 76%, but the specificity dropped to 37%. However, even when analyzed with the ITB G2 cut-off value, the predictive value for OS was still significant (p = 0.003, data not shown). Therefore, the prognostic impact of ITB was consistently valuable in DBDC, regardless of the high-level definition used.

In DBDCs, preoperative assessment of the T category was clinically difficult because tumor invasion depth is measured on postoperative histology. Although the stage of disease at DBDC presentation is known to be the most important prognostic factor [1], it is ironic that the disease stage cannot be evaluated at the time of diagnosis. Previous studies reported that ITB in CRC biopsies helped predict high-risk features indicative of poor prognosis prior to surgery [7, 8]. Zlobec et. al [7] investigated the prognostic value of ITB in preoperative biopsies of CRC patients with no neoadjuvant therapy: ITB^High, defined as more than 10 buds, consistent with our definition, predicted nodal and distant metastases. Rogers et. al evaluated the association of ITB in pretreatment rectal cancer biopsies with response to neoadjuvant chemoradiotherapy: ITB at the diagnosis of rectal cancer helped identify patients who would respond poorly to neoadjuvant chemoradiotherapy and those with a poor prognosis [8]. In this study, we found the prognostic utility of ITB in the T1 and/or N0 groups. Therefore, we postulated that ITB might preoperatively aid in predicting the prognosis, especially in early-stage DBDCs. However, our results had the limitation of being a retrospective study using resected specimens. Therefore, further multi-institutional studies using preoperative DBDC biopsy specimens are needed to obtain the robust prognostic utility of ITB in DBDCs.

In conclusion, high-grade buddings, in both PTB and ITB, were powerful predictive parameters for aggressive behavior and worse prognosis in patients with DBDC. ITB might be used as a surrogate marker in DBDC where PTB is difficult to assess. Specifically, in patients with stage I DBDCs, ITB can be an aid used to select high-risk features indicative of poor patient prognosis.

This work was presented in part at the European Society for Medical Oncology (ESMO) Congress 2023, Madrid, Spain, October 20th–24th, 2023.

This study protocol was reviewed and approved by the Institutional Review Board from two institutions (OC13SISI0162 from Incheon St. Mary’s Hospital and 2013-0527 from Asan Medical Center). The retrospective study has been granted an exemption from requiring written informed consent, which was approved by the Institutional Review Board from two institutions.

The authors have no conflicts of interest to declare.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. 2021R1A2C1003898, awarded to Sun-Young Jun) and by a Grant of Translational R&D Project through the Institute for Bio-Medical Convergence, Incheon St. Mary’s Hospital, the Catholic University of Korea (awarded to Sun-Young Jun). The sponsor of the study did not have any role in the study design or collection, analysis, and interpretation of data.

Sun-Young Jun conceived and designed the project, provided funding support, and wrote the manuscript. Sun-Young Jun and Seung-Mo Hong performed pathologic analysis, data collection, and database development. Soyeon An performed statistical analysis and data visualization. All the authors contributed to manuscript editing.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author upon reasonable request. Further inquiries can be directed to the corresponding author.