1 INTRODUCTION

Intrahepatic cholangiocarcinoma (ICC) is one of the primary hepatocellular carcinomas, second only to hepatocellular carcinoma in its incidence. ICC arises from the epithelial cells of either small intrahepatic ducts or large intrahepatic ducts near the bifurcation of intrahepatic ducts.¹ In both clinical situations, the pathological findings are usually classified as adenocarcinoma, although mixed hepatocellular cholangiocarcinoma also occurs, particularly in chronic liver disease. Despite its rarity, ICC is difficult to detect and treat and tends to be advanced at diagnosis.

Surgical resection is the only established option for treating ICCs, offering the best chance of cure.² However, because many patients with ICC have large, locally advanced tumors that require technically complex and challenging surgery, only about 20%–40% of patients with operable disease undergo surgical resection. In addition, the incidence of lymph node metastases (LNM) reportedly ranges from 17% to 62% and reflects a poor prognosis.^3-6 Despite LNM being a significant predictor of poor prognosis, the evidence of therapeutic benefit from lymph node dissection is insufficient and no consensus has been reached regarding whether lymphadenectomy should be routinely performed.

In terms of clinical implications, routine lymphadenectomy has been considered paramount for accurate staging and determination of prognosis and for guiding the postoperative treatment strategy.^{2, 7} Moreover, recent clinical studies have consistently shown the oncological efficacy of lymphadenectomy for ICC and have thus tended to support this approach.^{8, 9} However, clear evidence is lacking regarding the extent of lymphadenectomy required for accurate staging and the prognostic value of lymphadenectomy for each lymph node station. Moreover, ICCs invading the hilar structures tend to be associated with a higher incidence of LNM than peripherally located tumors.^10-12 Basing the extent of routine lymphadenectomy for accurate staging and treatment on tumor localization may thus be preferable.

The present study therefore aimed to investigate the survival benefit of lymphadenectomy among patients undergoing resection for ICC using a therapeutic index. We also evaluated the optimal extent of lymph node dissection for accurate staging and treatment by assessing tumor activity based on tumor localization.

2 METHODS

3 RESULTS

Overall patient background characteristics and differences between tumor locations are summarized in Table 1. The median duration of follow-up was 26.7 months interquartile range [IQR] 12.2–51.0 months). Median age at surgery was 70 years (IQR 63–76 years). The predominant tumor morphology was the mass-forming type (MF; n = 198), followed by the MF and periductal infiltrating type (PI; n = 42), the PI type (n = 30), and the intraductal growth type (n = 9). Regarding tumor location, all ICCs were distributed to the right peripheral lesion (n = 60), hilar lesion (n = 130), and left peripheral lesion (n = 89). Compared with peripheral lesions, hilar lesions required bile duct resection and vascular reconstruction, leading to a longer operation time and larger blood loss as surgical outcomes. The extent of lymph node dissection varied between centers. Of the 15 participating institutions, 12 institutions had performed lymphadenectomy beyond Level 1 in 80% of cases. The other three institutions also performed the same extended dissection in 40%–50% of cases. (Figure S1). Of the 279 patients who underwent lymphadenectomy, 109 patients (39%) showed histologically confirmed LNM. According to tumor location, the highest rates of both LNM and LNM beyond Level 1 were seen in the hilar lesion, followed by the left peripheral lesion and right peripheral lesion. The extent of LNM comprised Level 1-LNM (n = 31, 11%), Level 2-LNM (n = 44, 16%), and Level 3-LNM (n = 34, 12%). In the distribution of LNM, some overlap in cases was evident. All patients classified with Level 2-LNM were identified as showing LNM in Level 1. On the other hand, among patients classified into Level 3-LNM, two patients showed no evidence of positive nodes in Level 2, and one patient showed positive nodes in neither Level 1 nor Level 2 (Figure S2). Considering postoperative morbidity after lymphadenectomy, rates of severe complications (defined as Clavien–Dindo [CD] >grade II) were comparable between lymphadenectomies for Levels 1, 2, and 3. The rates of CD >grade II were 17% in Level 1, 19% in Level 2, and 18% in Level 3, respectively (data not shown, P = 0.168). However, CD grade II, represented by gastric stasis, was prominent in lymphadenectomy for Level 3, occurring in 19 cases for Level 3, only one case for Level 2, and no case for Level 1.

In risk analysis for LNM by using patient background and tumor factor, logistic regression analysis identified “component of PI” (odds ratio [OR] 2.90, P = 0.002), “preoperative CA19-9 level ≥ 118 U/ml” (OR 6.77, P < 0.001), “serosa invasion” (OR 1.81, P = 0.043), and “moderate or poor differentiation” (OR 4.51, P = 0.001) as independent risk factors for LNM (Table S1). An overview of LNM rates for each lymph node station and the therapeutic value of lymphadenectomy is provided in Table 2. Concerning TI for the overall cohort, 2-/5-y TI by lymphadenectomy were 12.8/7.2 for Level 1, 10.7/5.5 for Level 2, and 4.8/1.9 for Level 3. Subgroup analysis revealed that 5-y TI varied between tumor locations; the 5-y TI of lymphadenectomy in the hilar lesion was 4 for Level 1, 6.6 for Level 2, and 1.7 for Level 3. Moreover, the left peripheral lesion showed TIs of 7.5 for Level 1, 4.5 for Level 2, and 2.7 for Level 3. Particularly in the left peripheral lesion, incidences of metastasis in left gastric artery nodes and gastrocardiac nodes were 27.8% and 21.7%, respectively. Although the 2-y TI for these nodes was approximately 10, the 5-y TI was only 2.7. On the other hand, the right peripheral lesion contributed to 5-y TI only for Level 1 (Figure 1).

In survival analysis, median survival time (MST) and 1-/5-year overall survival rates after initial surgery were 52.4 months and 89.1%/45.3% for N0 disease, 19.1 months and 68.9%/27.1% for Level 1-LNM, 15.7  months and 61.2%/22.9% for Level 2-LNM, and 15.9 months, and 64.1%/7.3% for Level 3-LNM (P < 0.0001; Figure 2). Level 3-LNM showed the worst outcomes compared with the others. Conversely, the prognoses of Level 1-LNM and Level 2 were comparable. In subgroup analysis stratified by tumor location, the prognoses of LNM from Level 1 to Level 3 were comparable for the right and left peripheral lesion (Figure S3).

Concerning recurrent outcomes after surgery, N0 disease showed longer recurrence-free survival (RFS) than did Level 1-LNM, Level 2-LNM, and Level 3-LNM: median RFS was 29.8 months for N0 disease, 8.3 months for Level 1-LNM, 9.2 months for Level 2, and 6 months for Level 3 (P < 0.001, Table 3). Furthermore, the highest early recurrence rate was seen for Level 3-LNM, followed by Level 2-LNM, Level 1-LNM, and N0 disease (P < 0.001, Figure 3). The proportion of extrahepatic spread as recurrence increased with the extent of LNM (P = 0.083, Table 3). Most patients with Level 2- or Level 3-LNM showed both early recurrence within 1 year after surgery and extrahepatic recurrence. Level 2- and Level 3-LNM showed multisite recurrence, with an average of 1.3 recurrent sites in Level 3-LNM, 1.2 recurrent sites in Level 2-LNM, 1.2 recurrent sites in Level 1-LNM, and 0.7 recurrent sites in N0 disease (P < 0.001, Table 3). Approximately 30% of recurrent patients with Level 2- or Level 3-LNM showed 2–4 sites of recurrence, including remnant liver, lungs, peritoneum/pleura, and LNM. Reflecting these results, surgical resection rates for recurrent disease were 15% with N0 disease, 13% with Level 1-LNM, 5% with Level 2-LNM, and 3% with Level 3-LNM. The proportion of best supportive care was increased by the extent of LNM. Survival curves for patients were categorized according to the number of sites of recurrence (Figure 4). In the Cox proportional hazard model, “Tumor size≥5 cm” (hazard ratio [HR] 1.48, P = 0.025), “Hilar lesion” (HR 1.55, P = 0.08), “Multinodular” (HR 1.46, P = 0.048), “CA19-9 ≥ 118 U/ml” (HR 1.71, P = 0.004), and “Extent of LNM” (HR 1.82–2.89, P < 0.01) were identified as independent poor prognostic factors (Table 4). In particular, the prognostic impacts of the extent of Level 1-, Level 2-, and Level 3-LNM were well categorized.

4 DISCUSSION

Routine lymphadenectomy for ICC has been controversial because of the lack of clinical evidence. Nevertheless, recent clinical studies have supported lymphadenectomy for oncological treatment and staging.^{8, 9} Thus, in current trends, the proportion of patients undergoing lymph node dissection for ICC has increased year by year, particularly in Western countries.¹⁷ However, specifics of the scope and methods of lymphadenectomy have yet to be defined. The American Joint Committee on Cancer (AJCC) recommends harvesting at least six lymph nodes to ensure accurate staging.¹⁸ International collaborative studies were concordant with this recommendation.^{19, 20} The National Comprehensive Cancer Network guidelines also recommend regional lymphadenectomy in the hilar region of the liver, while AJCC guidelines recommend a broader level, depending on tumor location.^{18, 21} According to the AJCC guidelines, lymphadenectomy should be performed beyond the hepato-duodenal ligament, depending on the localization of the tumor. However, this proposal lacks sufficient evidence-based data and is instead based on expert recommendations. The present study aimed to evaluate the efficacy and significance of lymph node dissection for ICC and to define the optimal extent of lymph node dissection according to tumor localization. With constraints such as the relative rarity of ICC and the commonly accepted surgical strategy of lymphadenectomy, establishing a randomized controlled study would be invaluable, but unrealistic. Some previous studies have reported the therapeutic benefit of lymphadenectomy for ICC.^{8, 9, 22} However, those studies were mainly retrospective studies using simulation analysis, propensity score matching, and inverse probability of treatment-weighting approaches with propensity scores. The present study focused on TI as a promising index with proven implications in gastric cancer, where the efficacy of lymph node dissection is well established.^{16, 23} The significance of TI has been demonstrated more recently for various gastrointestinal cancers, including colorectal cancer, esophageal cancer, and pancreatic cancer.^24-27 Sahara et al²⁸ reported on the significance of TI for ICC, and identified which patients would benefit from lymphadenectomy.

Contrasting with their report, we focused on tumor location and set the valuable index to 5 y. This was because ICC has different aspects of lymph node metastasis, depending on tumor location^{11, 22, 29, 30} and we wished to evaluate the efficacy of lymphadenectomy as a benchmark for other carcinomas for which lymphadenectomy has been established as the standard surgical procedure. Furthermore, we adopted simple zoning of each lymph node station around the liver with the lymphatic drainage route and practical use of lymphadenectomy. According to these classifications of lymph node, the implications of Level 2 include more accurate staging than Level 1,¹⁸ while those of Level 3 include a left pathway or cardinal route as a higher likelihood of lymphatic basin, particularly for left lateral ICCs.^30-32

In our results, the incidence of LNM heavily depended on tumor location: hilar lesions showed the highest LNM potential, followed by left peripheral and right peripheral lesions. There was no doubt that Level 1 had the most frequent nodes as a primary lymph node station, and Level 2 could represent possible metastatic nodes from the left peripheral and hilar regions. On the other hand, the right peripheral region seldom showed lymphatic spread to Level 3. In our cohort, very few Level 2-LNM or Level 3-LNM were seen without Level 1-LNM. In other words, Level 2- or Level 3-LNM is usually accompanied by Level 1-LNM. Therefore, Level 3 as a left pathway or cardinal route should be considered as an independent lymphatic route, particularly for left lateral ICCs, but in terms of therapeutic efficacy and staging significance, should be regarded as second only to Level 1 and Level 2.

Concerning therapeutic value, the 5-year TI of lymphadenectomy was 7.2 for Level 1, 5.5 for Level 2, and 1.9 for Level 3. According to tumor localization, the hilar lesion showed 5-year TI > 5.0 in Level 1 and Level 2. On the other hand, bilateral peripheral lesions were only seen in Level 1. Based on these results, lymphadenectomy for Levels 1 and 2 could exert a positive impact on prognosis for hilar lesions. As well as this, lymphadenectomy for Level 1 could have a positive impact on peripheral lesions. To justify the indications for lymphadenectomy, future work should examine where best to set the cut-off for TI. Although the efficacy would be limited, lymphadenectomy for Level 3 in the left peripheral region is worth considering.

Adjuvant therapy has recently been considered more closely for further achieving improvements of the surgical prognosis for ICC. While the clinical benefits of adjuvant treatment for ICC remain unclear, the potential survival benefits of adjuvant chemotherapy could be associated with tumor subgroups, such as the presence of LNM.³³ From this perspective, accurate staging through lymphadenectomy is indispensable. Regarding the precise stage for estimating the degree of cancer spread, Level 1- to 3-LNM could reflect the degree of cancer spread. Multisite recurrence, including to the remnant liver, lymph nodes, and other organs, could relate to lymphatic spread. Expanding the range of lymphadenectomy could thus lead to more accurate staging.

Nonetheless, it must be noted that lymphadenectomy is also associated with a specific risk of postoperative morbidity. Postoperative morbidity would possibly delay the introduction of adjuvant chemotherapy. Extending lymphadenectomy to Level 3 could increase the occurrence of CD Grade II complications, among which gastric stasis is prominent. These patients required longer hospitalization and delayed introduction of adjuvant chemotherapy. Preemptive pyloroplasty may help prevent this situation.³⁴ However, the indications for lymphadenectomy for Level 3 disease should be considered based on the balance between therapeutic effects and the significance of staging and its disadvantages. Consequently, determination of the extent of, and indications for, lymph node dissection is essential, taking into account the therapeutic and diagnostic implications according to the location of the ICC.

Some limitations to this study should be kept in mind. This analysis focused on TI to confirm the clinical implications of lymphadenectomy for ICC. However, nodal status may migrate in cases where the extent of routine lymphadenectomy would be different in a retrospective, multicenter collection of cases. Therefore, even though more than 80% of the patients in this study had undergone Level 2 or higher lymphadenectomy, it is important to note that the patient group is comprised of a heterogeneous population with an extent of dissection. Furthermore, the actual TIs were also indicated in Level 2 and 3 lymph nodes. It should be noted, however, that TI could be calculated based on the survival rate of very rare positive metastases, as seen in the peripheral-type of pancreatic retroperitoneal lymph nodes.

These results may be difficult to accept because of previous reports referring to dismal results from lymphadenectomy. However, our results suggest that a certain number of patients will gain therapeutic value from lymphadenectomy significant enough to determine the exact extent of ICC. Moreover, the classification of lymph node level could contribute to determination of the extent of lymphadenectomy. We report the therapeutic value and clinical implications of lymphadenectomy for ICC differently from previous reports. Further, tumor location could offer a significant determinant for the priority of lymphadenectomy.

5 CONCLUSIONS

Lymphadenectomy achieved by liver resection with curative intent should be a minimal requirement for the multimodal treatment of ICC. Furthermore, the implications and extent of lymphadenectomy for ICC should be determined according to the tumor location and preoperative risk of LNM. In peripheral type ICC, the prognostic benefit of lymphadenectomy would be limited and dissection beyond level 1 should be eliminated, while in hilar-type, lymphadenectomy up to level 2 would be necessary.

AUTHOR CONTRIBUTIONS

Yuzo Umeda: Study design, data analysis, and writing the article. Kosei Takagi, Tomokazu Fuji, and Tatsuo Matsuda: Data analysis. Toru Kojima, Daisuke Satoh, Masayoshi Hioki, Yoshikatsu Endo, Masaru Inagaki, Masahiro Oishi, Takahito Yagi, and Toshiyoshi Fujiwara: Data collection.

FUNDING INFORMATION

Financial support was received from the Japan Society for the Promotion of Science (grant number 22 K08775 to Yuzo UMEDA).

CONFLICT OF INTEREST

The authors declare no conflicts of interest for this article.

ETHICS STATEMENT

Approval of the research protocol: This study conformed to the Declaration of Helsinki on Human Research Ethics standards and was approved by the Okayama University Hospital Institutional Ethics Board (approval no. 2208–016).

Informed Consent: The need for written, informed consent was waived by the Okayama University Hospital Institutional Ethics Board because of the retrospective design of this study.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: N/A.