MATERIALS AND METHODS
Study design
A prospectively maintained institutional database of oncologic liver resections was analyzed retrospectively. The study included all patients who underwent liver resections from 1 January 2011 to 31 December 2021 at a tertiary cancer care hospital in Mumbai, India.
Primary outcome analyzed was disease-free survival (DFS), while secondary outcomes were OS, postoperative complications, and recurrence patterns. Benign cases were excluded for all survival calculations.
Major hepatectomy was defined as the removal of three or more contiguous liver segments, as per the definition proposed by Bismuth and Chiche [6]. All treatment decisions were made by a multidisciplinary board (comprising a Hepatobiliary surgeon, a medical oncologist, a radiation oncologist, and an interventional radiologist), based on guidelines available at the time. All patients had drain bilirubin measured on day 3 and 5 as per institutional policy. Postoperative bile leak was defined as per the International Study Group of Liver Surgery (ISGLS) definition [7], and clinically significant bile leaks were defined as ISGLS grade B or C leaks. DFS was defined as time from surgery to disease recurrence or death due to any cause, or last known follow up, whichever was earlier. Overall survival (OS) was defined as time from surgery to death due to any cause, or last known follow up, whichever was earlier.
All patients received standard postoperative care as per the Enhanced Recovery After Surgery (ERAS) protocol, and were discharged when they were ambulant, tolerating an adequate oral diet, had adequately controlled pain, and were free of uncontrolled postoperative complications. Follow up was scheduled every 3 months for the first 2 years, and 6 monthly till 5 years, with cross sectional imaging every 6 months for the first 2 years, and yearly after that.
This study was carried out in concordance with the ethical standards set by the institutional ethics committee and being a retrospective review of previously collected data, without any patient identifiers being disclosed, a waiver for the need of patient consent was granted.
Statistical analysis
All statistical analyses were performed using IBM SPSS Statistics version 28.0. Continuous variables were represented using median with interquartile range, and comparison was made using a Mann–Whitney U-test. Categorical data was represented in percentages, and analyzed using a Pearson chi-square test. OS and DFS were assessed using Kaplan–Meier method, and outcomes were compared using the log–rank test. Follow up duration was assessed using the reverse Kaplan–Meier method. Data was further stratified as per the histology.
RESULTS
Demographic details
The subjects were eight hundred sixty-two patients who underwent liver resections between January 2011 to November 2021, with 306 (35.5%) hepatocellular carcinoma (HCC), 212 (24.6%) metastatic colorectal cancers, and 111 (12.9%) cholangiocarcinomas (69 intrahepatic cholangiocarcinomas, 42 hilar cholangiocarcinomas), as summarized in Table 1. The incidence of preoperative jaundice and biliary drainage was more common in patients who developed a PHBL (preoperative biliary drainage in 11.0% vs. 3.2%, p < 0.001), and there was a higher proportion of cholangiocarcinomas in the PHBL group (23.3% vs. 10.7%, p < 0.001).
Operative and long-term outcomes
One hundred forty-six (16.9%) patients had a postoperative bile leak, with 60 (7.0%) having a clinically significant bile leak (ISGLS grade B or C) with 34 (3.9%) patients experiencing postoperative mortality. Four hundred forty-nine (52.1%) patients underwent a major hepatic resection (3 or more segments resected), 58 (6.7%) patients a bilio-enteric drainage, and 11 (1.3%) patients a vascular reconstruction. Table 2 summarizes the operative details and postoperative outcomes.
Long-term outcomes
The median follow up was 14.8 (6.6−35.4) months. Supplementary Fig. 1 of the supplementary information depicts the survival outcomes for the entire cohort, excluding benign histologies. There was no significant difference in either OS (18 month OS 79.4% vs. 61.9%, p = 0.695) or DFS (median DFS 19.4 months vs. 22.6 months, p = 0.285). On subgroup analysis as per histology, there was no significant difference in either OS or DFS in patients with HCC (18 month OS 80.2% vs. 86.8%, p = 0.63 and median DFS 26.2 months vs. 25.3 months, p = 0.980) or metastatic colorectal cancer (3 year OS 72.4% vs. 59.2%, p = 0.507 and median DFS 13.1 months vs. 16.7 months, p = 0.361), as shown in Fig. 1.
Survival outcomes only showed a significant difference in patients with cholangiocarcinomas (n = 111), and so further analysis was restricted to this cohort of patients.
Subgroup analysis of patients with cholangiocarcinoma
Table 3 and 4 depict the demographic details and outcomes, respectively, for cholangiocarcinoma patients. Of the 34 patients who developed PHBL, 12/34 had a Grade A leak, 19/34 had a Grade B PHBL, and 3/34 had a Grade C PHBL. Of the 34 patients who developed a PHBL, 26 (76.5%) did not receive any adjuvant therapy, and only 5 (14.7%) received adjuvant therapy within 4 weeks of surgery, as opposed to patients with no PHBL, of whom 35.1% received adjuvant therapy. As expected, the postoperative stay was higher in patients who developed a bile leak. Rate of major hepatectomy 97.0% vs. 76.6% (p = 0.008) and blood loss (median blood loss 2,600 mL vs. 1,800 mL, p = 0.026) were also significantly higher in the group of patients who developed bile leaks. Fig. 2 depicts the OS and DFS for cholangiocarcinoma patients. Patients with cholangiocarcinoma who developed a bile leak had a significantly worse DFS (median DFS 9.9 months vs. 24.9 months, p = 0.013). There was a trend towards worse OS in patients who developed bile leak, though this was not statistically significant (median OS was not reached in either group, 2 year survival 61.9% vs. 85.9%, p = 0.09). Cox regression was performed to predict factors associated with a DFS detriment, as shown in Table 5. This was primarily done to determine whether the DFS detriment seen with bile leaks was due to the higher proportion of major hepatectomies and biliary reconstructions in this cohort. On univariate analysis, development of bile leak, nodal positivity, and margin positivity were associated with shortening of DFS, and this persisted on multiple regression analysis:
Node positivity (hazard ratio [HR]: 2.482, p = 0.033), margin positivity (HR: 2.65, p = 0.021), development of PHBL (HR: 1.896, p = 0.033).
DISCUSSION
The current study reported an incidence of PHBL of 17%, which is higher than that reported in the literature [1,2], but can be explained by the higher number of major hepatectomies performed along with the routine measurement of drain bilirubin on day 3 after surgery. This results in a large number of otherwise asymptomatic leaks being diagnosed, which is reflected by an incidence of clinically significant bile leaks of only 7%, which compares well with the incidence of 3%−10% reported in the literature [1-3]. A previous study had identified blood loss, bilio-enteric anastomosis, and extended resections as independent risk factors for PHBL [8], which is reflected in our study with the PHBL group having a higher median blood loss, and a higher proportion of major hepatectomies and bilio-enteric anastomosis. The PHBL group also had a higher proportion of cholangiocarcinomas (23.3% vs. 10.7%, p < 0.001) and lower proportion of colorectal metastasis (15.1% vs. 26.5%, p < 0.001), likely due to the cholangiocarcinomas more likely requiring extended resections with higher blood loss, and the extensive dissection of the biliary tract and hilar dissection required in cholangiocarcinomas increasing the risk of bile leaks. Other studies have also found this, with PHBL occurring more frequently in cholangiocarcinomas than in HCC and other liver tumors [3,4].
The only sub-group of patients in whom development of PHBL impacted survival was cholangiocarcinoma patients. The adverse impact on DFS in cholangiocarcinomas could be related to the delay in adjuvant therapy delivery, with 26/34 (76%) patients not receiving chemotherapy, and only 5/34 (15%) of the patients received chemotherapy within 4 weeks of surgery. A National Cancer Database (NCDB) analysis by Parsons et al. [9] showed that delays in adjuvant therapy are associated with detriments in survival in patients with cholangiocarcinoma, with a significant OS detriment if treatment was initiated more than 59 days from surgery. Delay of chemotherapy is not as important in other histologies, with there being no role of adjuvant therapy in HCC. A factor that could be investigated is whether earlier intervention can reduce the delay in the administration of chemotherapy. Standard practice is to manage small bile leaks conservatively, and only consider interventions such as stent placement in major bile leaks, or those that fail to respond to conservative management. Whether earlier intervention could result in earlier administration of chemotherapy is an issue that should be investigated.
The inflammatory milieu promoted by bile leaks could also play a role in worsening outcomes. Wang et al. [4] found that PHBL is associated with increased levels of inflammatory cytokines, which have been shown to worsen prognosis in other cancers [10,11]. Our study found that while PHBL is associated with a detriment in survival, most of this detriment was driven by clinically significant bile leaks, and Grade A PHBL was not associated with any survival detriment (median DFS 18.6 months vs. 24.9 months, p = 0.647). This is similar to the findings reported by Wang et al. [4], which found that while long-term bile leaks persisting > 4 weeks adversely impacted DFS, short-term leaks did not. While there was a trend toward OS detriment in the PHBL group, the short follow up period in our study precludes any meaningful conclusion from being drawn.
Numerous studies have proposed many validated scores to predict PHBL, and there are well defined preoperative factors that can predict the risk of PHBL [8,12]. Given the mounting evidence that points to PHBL being an adverse prognostic factor, future studies should be directed toward addressing this issue by either intensifying adjuvant therapy in patients who develop PHBL, or by delivering systemic therapy in a neoadjuvant fashion to patients at high risk of developing PHBL. The role that the inflammatory response plays in this survival detriment also warrants further investigation.
In our study, the impact of PHBL on survival was restricted only to patients with cholangiocarcinoma. Other studies have also shown that PHBL does not impact long-term survival in patients with HCC or colorectal liver metastasis [13,14], and the reason why the survival detriment of PHBL is limited to only cholangiocarcinomas is unclear, but could be related to several factors. Firstly, systemic therapy plays a major role in the management of cholangiocarcinomas, unlike in HCC. Secondly, biliary leaks from tumors arising from within the bile ducts could hypothetically lead to tumor seeding from exfoliated cancer cells in the biliary tract, which when combined with the inflammatory milieu provided by the bile leak, provides an ideal environment for tumor seeding.
There are a few limitations to this study. Being retrospective in nature, there is an inherent selection bias introduced, and the fact that the study population consisted of patients treated more than a decade apart has led to numerous differences in the treatment given. For example, currently used adjuvant agents, such as S−1, were not available initially. Further, standardization of surgical procedure and widespread adoption of the ERAS protocol occurred somewhere during the middle of the study period, leading to improvement of surgical outcomes and reduction in hospital stay [15]. Also, despite a sample size of 862 patients, there were only 111 cholangiocarcinoma patients; this limited sample size makes delving into the numerous competing factors involved in PHBL difficult. For example, only 12 cholangiocarcinoma patients developed a Grade A PHBL, and thus the impact that Grade A PHBL has on long-term outcomes cannot be adequately assessed. Lastly, the limited follow up duration precludes adequate assessment of OS.
In conclusion, the development of a PHBL after liver resection is an adverse prognostic factor in patients with cholangiocarcinoma, with patients who develop a PHBL having a significantly worse DFS than those who do not, but this survival detriment does not occur in patients with HCC or colorectal liver metastasis. Including PHBL as an adverse prognostic factor in patients with cholangiocarcinoma and investigating methods to mitigate the survival detriment it poses need to be investigated.