4. Discussion
This study represents our retrospective experience of surgical treatments for perihilar cholangiocarcinomas in the last decade. The primary endpoint was to investigate the role of lymph node metastasis for patient survival after complete tumor resection.
In cases of lymph node metastases, the outcome was severely compromised and comparable to patients with an incomplete resection (R1). Usually, successful surgical treatment in carcinoma patients is defined as carcinoma-free resection margins. However, in our cohort of patients, the resection margin alone poorly differentiated the long-term survival and was significantly influenced by the lymph node status of the patients. This is why we defined a successful surgical treatment as free resection margin in patients with negative lymph node status. Such patients show the true benefit from surgical treatment and have a clinically relevant long-term prognosis, as documented in
Figure 1. This was demonstrated in other patient cohorts as well: Benzing et al. showed that the median survival is not different for patients with lymph node metastases in cases of complete (R0) or incomplete (R1) resection [
12]. This underlines the importance of a systematic locoregional lymphadenectomy in patients with perihilar cholangiocarcinoma for a correct staging and individual risk assessment. This fosters the conclusion of the BILCAP Study to apply adjuvant chemotherapy following surgery for biliary tract cancer, at least in all patients with positive lymph nodes [
7]. However, aggressive lymphadenectomy does not go without a price to pay:
The example of pancreatic cancer seems appropriate to judge the meaning of an extended lymphadenectomy in oncologic surgery of the upper GI. While a regional lymphadenectomy improves the staging of the disease, resection of secondary or third level lymph nodes would only improve survival without other microscopic disease. Unfortunately, only a minority of patients presumably develop microscopic disease in second or third level lymph nodes without other systemic occurrences. As such, only a small minority of patients will recur with isolated unresected second level metastatic lymph nodes. It has been concluded that removing negative lymph nodes in secondary level lymph nodes will not improve disease control. On the other hand, a higher rate of intra- and postoperative complications have been observed after extended lymphadenectomy in different oncologic entities [
13,
14,
15]. Based on such extrapolation from other entities, it seems appropriate to perform a systematic and careful lymphadenectomy in the primary nodes for proper staging and perform adjuvant therapies in case of any suspicion of microscopic systemic disease. A recent multi-institutional study suggested that the optimal minimal number of resected lymph nodes in perihilar cholangiocarcinoma is four nodes [
16], which further highlights a mandatory lymphadenectomy of the primary nodes, without excessive extension of surgery resulting in more complications. Other studies reported a significant influence of the lymph node ratio after resection of perihilar cholangiocarcinoma [
17,
18], which was not observed in our cohort of patients (data not shown).
From our perspective, the presented data are in accordance with the literature, suggesting a standardized lymphadenectomy of the primary nodes resulting in four lymph nodes in every patient undergoing surgery for perihilar cholangiocarcinoma.
Due to this ongoing discussion, we aimed as a secondary endpoint for the prediction of the lymph node status from preoperatively available parameters, which is of utmost interest from a clinical point of view. Our analysis demonstrated that such predictions are hardly possible: The only independent parameters predictive for this endpoint were the Bismuth-Corlette stage and the histopathological grading. While both were assessable in several preoperative cases, both could not be influenced and represent advanced disease.
On the other hand, this underlines the relevance of precise preoperative diagnostics using cholangiography and concomitant biopsy and should be the substantial basis to inform and advise the patient of the correct expectation of what is achievable by surgery.
Future work might establish neoadjuvant treatment concepts for such patients to improve outcomes after surgery. Unfortunately, successful neoadjuvant concepts are missing so far.
It should be of interest that the best diagnostic tool in cases of biliary tract cancer is still under debate: While some authors suggest percutaneous biliary drainage (PTBD) for superior delineation of tumor involvement of the biliary tree, as well as superior anatomic stent placement [
19], it carries the disadvantage of external drains. On the other hand, ERCP is associated with higher rates of stent misplacement, resulting in additional procedures and time delays in surgical treatment. Unfortunately, a multicenter, randomized controlled trial from the Netherlands [
20] investigating the best diagnostic tool prematurely ended with increased mortality in the PTBD group compared with the ERCP group. However, several limitations of this study warrant cautious interpretation. Overall and preoperative complications were similar in both groups. Moreover, other studies describe higher rates of infections after ERCP compared with PTBD in patients with advanced biliary tract cancer [
21].
In this context, it should be pointed out that preoperative biliary drainage is still under debate, and some data demonstrated that drainage does not affect overall mortality in jaundiced patients [
22]. Therefore, non-invasive methods, such as MRCP alone, might be considered for diagnostic workups. However, based on the present data with the tumor grading as an independent predictor of lymph node metastasis and the patient’s overall survival after surgery, a biopsy giving some evidence of the tumor’s characteristics might be of interest preoperatively.
Last, our secondary endpoints included predictive factors of overall patient survival. Regarding the overall survival, we found, as expected, that patients with advanced disease, as represented by lymph node metastasis and incomplete resection margins, did worse. This proves that only patients with a limited disease can be transferred into curative strategies. The 1-, 3-, and 5-year survival rates for curatively resected patients in this study were 84.5%, 38.7%, and 20%. With a 1-year survival rate of 87.7%, patients in this study had an above-average survival rate compared with resected patients in other studies. The 3- and 5-year survival, on the other hand, can be classified as average in the present study situation. The following 1-, 3-, and 5-year survival rates were observed in current studies for curatively resected patients: 40–70%/27–56%/10–42% [
23,
24,
25,
26,
27].
Additionally, the preoperative level of bilirubin was identified as an independent predictor of overall patient survival, as observed by others [
28]. We want to point out that most of our patients are treated with preoperative bile duct stenting during ERCP. This suggests that the preoperative level of bilirubin may effectively reflect the severity of the disease and, therefore, of advanced and complex cases, which, despite preoperative stent insertion, have an incomplete reduction of cholestasis or only partially successful endoscopic drainage (e.g., only one-sided drainage). Intraoperative blood transfusion was also an independent predictor of overall survival after censoring for in-house mortality. This should be seen in the context of the low transfusion rate of only 10% of the population under investigation. Of course, the intraoperative transfusion of RBCs might resemble more complex surgical procedures, and therefore advanced disease. However, other explanations, especially in the context of the overall survival, include transfusion-related immunomodulation (TRIM) leading to diminished immune surveillance and the elusion of micrometastases. Interestingly, the promotion of growth in cancer cells via blood transfusions has been shown in animal models [
29]. However, the exact mechanism underlying TRIM remains unclear [
30]. Several cell lines (natural killer cells, T cells, suppressor T cells, macrophages, and monocytes) have been suggested to be altered in count and function [
31]. This result aligns with other investigations, identifying blood transfusions as an independent predictor of survival in hepatobiliary surgery [
32].
Limitations of our study include the monocentric retrospective design and the limited number of patients. The retrospective aspect may introduce selection bias and misclassification or information bias. When relying on individual recall of former exposure to risk variables, the recall may be inaccurate and subject to biases. It can be tough to accurately compare the exposed and the non-exposed. Another relevant limitation of the study is the lack of information regarding applied adjuvant therapies. Since reliable data on that matter were only collected within our hospital postoperatively, disease-specific survival could not be collected in all cases and therefore had to be reported as overall survival. Additionally, one-third of the existing cohort had an ASA classification of three at the time of surgery, suggesting relevant comorbidities with an impact on overall survival. On the other hand, it can be assumed that other study cohorts reporting on cohorts of the same average age at diagnosis underly the same bias, so there is still comparability. However, there are also strengths of our study. First, all patients were treated by a few different surgeons and anesthesiologists. All underwent standardized operations and anesthesia procedures, and all patients were followed at a dedicated ICU. Moreover, a cohort of 107 patients seems large enough to conclude relevance to treating a rare disease in a medical context.
Our results demonstrate the necessity and feasibility of surgical treatment in Klatskin carcinoma, with surgery as the only option for a curative treatment concept. However, a remarkably high number of patients were sent for surgery and found unresectable upon surgical exploration. The treatment of Klatskin carcinomas remains challenging due to the aggressiveness of the disease, its late diagnosis, and its complex anatomical location.
In conclusion, our data demonstrate the impact of resection margins in combination with the lymph node status for the survival of patients undergoing surgery for Klatskin tumors. Unfortunately, a good prediction of this subgroup of patients using preoperatively known parameters is limited.