Preoperative liver optimization

Patients were considered jaundiced when clinical jaundice was present upon initial examination and confirmed by elevated serum bilirubin level (>2.0 mg/dL). Of the 47 jaundiced GBC patients who underwent curative intent procedures, 19 underwent preoperative biliary drainage (PBD). Percutaneous transhepatic biliary drainage (PTBD) was performed in 10 patients (10/19; 52.6%). Endoscopic biliary drainage was performed in 9 patients (9/19; 47.4%). This intervention was performed approximately a week before hepatectomy based on the revaluation of hepatic function and the selection of the surgeon. No patient underwent preoperative portal vein embolization.

Operative procedures

Our center’s surgical policy for GBC involves radical surgery. For radical surgery, a partial hepatectomy with en bloc resection of GB and a dissection of regional lymph nodes (lymph nodes along the hepatoduodenal ligament and common HA and behind the pancreatic head) were routinely conducted (Figure 1). Partial hepatectomy includes extended right/left hepatectomy, right trisectionectomy, or wedge resection with a 2 cm margin (including segments IVb/V). Even advanced GBC was considered a candidate for resection as long as it could be anatomically removed. Extrahepatic bile duct resection was performed when the preoperative diagnostic images showed a tumor affecting the extrahepatic bile duct or when a tumor was significantly close to or invaded the extrahepatic bile duct upon intraoperative inspection (Figure 2).

Macroscopically involved adjacent organs were resected en bloc using major hepatectomy, pancreatoduodenectomy, partial gastrectomy, partial duodenal resection, partial colon resection, and/or portal vein/hepatic arterial resection and reconstruction as long as R0 resection was expected.

Surgeons assisted pathologists to correctly identify resection margins during the preparation of sections in fixed specimens. Surgical resection was considered to have curative intent (R0 or R1) when the whole tumor was resected, such that no macroscopically residual tumor could be detected. Stage grouping was performed according to the pTNM classification system of UICC, 7th edition [10].

Adjuvant therapy was given to 28 patients: intraoperative chemotherapy in 18 patients, postoperative chemotherapy in 15 patients (including a combination of intraoperative and postoperative chemotherapy in 13), postoperative radiotherapy in 15 patients, and a combination of chemotherapy and radiotherapy in 7 patients.

Statistical analysis

The overall survival was measured from the date of operation to death, including deaths caused by cancer or other causes, or until the last day of follow-up. The two groups were compared using Student’s t test for parametric data and the Mann-Whitney U test for non-parametric data. The Chi-square test was used for categorical data. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank test. Cox regression analysis was performed to determine which factor is the best prognostic determinant. Statistical significance was considered at p < 0.05. Statistical analyses were performed using SPSS Version 17.0 for Windows (SPSS, Inc., Chicago, IL, USA).

Demographic data

Surgical procedures

Table 1 summarizes the surgical procedures. In this study, "major hepatectomy" indicates right or left hepatectomy, extended right or left hepatectomy, or right or left trisegmentectomy while "minor hepatectomy" indicates segmental resection or less. Parenchymal transection was performed under HA and portal vein clamping for 15 min at 5 min intervals. Hepatectomy was performed in all 192 patients. Eight patients (8.5%) underwent major hepatectomy, including one patient with combined caudate lobe resection, three patients with partial portal vein (PV) resection, and two patients with HA resection and reconstruction. The following combined resections of other organs were performed in 15 patients: pancreatoduodenectomy (n = 2), wedge duodenal resection (n = 1), segmental colon resection (n = 1), partial kidney resection (n = 1), and partial gastrectomy (n = 10).

All patients underwent en bloc dissection of the regional lymph nodes (lymph nodes along the hepatoduodenal ligament and common hepatic artery and behind the pancreatic head). Morbidity was significantly lower in the non-jaundiced patients than in the jaundiced patients (12.4% vs. 34.0%, p = 0.001). Compared with the non-jaundiced patients, the jaundiced patients had significantly longer operative times (p < 0.001) and more intra-operative bleeding (p = 0.001), advanced T category (p = 0.019), common bile duct resections (93.6% vs. 8.3%, p < 0.001), extensive pathologic extrahepatic bile duct invasion (pEBI) (80.9% vs. 8.3%, p = 0.001), and frequent CRAOs (23.4% vs. 2.8%, p = 0.001) (Table 1). However, no significant differences in major hepatectomies and R0 resection were observed between the jaundiced and non-jaundiced patients (8.5% vs. 2.8%, p = 0.590 and 76.6% vs. 87.6%, p = 0.068).

Mortality and morbidity in 192 GBC patients who underwent curative intent procedures (Table 2)

The postoperative mortality and morbidity rates of the 192 GBC patients were 3.1% (n = 6) and 17.7% (n = 34), respectively. Postoperative complications were graded I in 2 patients, II in 12, IIIa in 11, IIIb in 2, IVa in 1, and V in 6 patients using Clavien–Dindo classification [11]. Considering the lack of a significant difference in mortality between the jaundiced patients and non-jaundiced patients (p = 0.141), we focused on morbidity. Univariate analysis showed that the risk factors for morbidity were preoperative jaundice (34.0% vs. 12.4%, p = 0.034) and intraoperative blood transfusion (34.1% vs. 12.8%, p = 0.056). Multivariate analysis revealed that preoperative jaundice was the only independent predictor of postoperative morbidity.

Mortality and morbidity in 47 jaundiced GBC patients (Table 3)

The postoperative mortality and morbidity rates in the jaundiced patients were 6.4% (n = 3) and 34.0% (n = 16), respectively. The causes of death were acute liver failure (n = 1) followed by renal failure; intra-abdominal bleeding (n = 1); and sepsis with multiorgan failure (n = 1).

Postsurgical complications were detected in 16 jaundiced patients. The most frequent complications were intra-abdominal abscesses (n = 9), biliary leakage (n = 2), intra-abdominal bleeding (n = 3), aspiration pneumonia (n = 1), and liver failure (n = 1). These complications required an invasive procedure in 13 patients (reoperation: n = 2, ultrasound guided drainage: n = 11). The average postoperative hospital stay of the jaundiced patients was 19.7 d (range: 4-85 d), which was longer than that of the non-jaundiced patients (p < 0.001). Univariate analysis identified no risk factor for postoperative morbidity among the jaundiced patients.

Survival and risk factors in 192 GBC patients who underwent curative intent procedures (Tables 4 and 5)

Overall, the three- and five-year survival rates and median survival time for the 192 patients were 35.3%, 28.1%, and 37.0 months, respectively (Figure 3). Survival curves for the 192 patients, grouped according to preoperative jaundice status, are shown in Figure 4. The five-year survival rate and median survival time were 6.0% and 14.0 months for the 47 jaundiced patients, respectively, and 36.0% and 43.0 months for the 145 non-jaundiced patients, respectively. The jaundiced patients had significantly lower survival rates than the non-jaundiced patients (p < 0.001).

Univariate and multivariate analyses were performed on the 192 GBC patients who received surgical resection with curative intent to identify the factors that influence long-term survival (Tables 4 and 5). Univariate analysis revealed that the significant risk factors of survival were age (p = 0.012), preoperative jaundice (p < 0.001), curative resection (p < 0.001), tumor location (p < 0.001), pT factor (p < 0.001), lymph node metastasis (p < 0.001), hepatic invasion (p < 0.001), CRAO (p < 0.001), combined portal vein/hepatic artery resection (p = 0.041), and intraoperative blood transfusion (p = 0.011) (Tables 4). Multivariate analysis was performed to determine which univariate prognostic relationships are independent predictive factors. Lymph node metastasis and tumors at the gallbladder neck were the significant risk factors of poor prognosis in this analysis (Table 5).The patients who underwent CRAO had significantly lower survival rates than those who did not undergo the procedure (p < 0.001). The patients who underwent R0 resection had a higher five-year survival rate than those who underwent R1 resection (p < 0.001). However, multivariate analysis revealed that neither CRAO nor R1 resection was an independent predictor of poor prognosis. The jaundiced patients had significantly lower survival rates than the non-jaundiced patients (p < 0.001) (Figure 4). However, multivariate analysis demonstrated that preoperative jaundice was not a significant risk factor (p = 0.295).

Survival and risk factors in 47 jaundiced GBC patients (Table 6)

The one- and three-year survival rates and median survival time of the 47 jaundiced patients were 37.8%, 6.0%, and 14.0 months, respectively. Univariate analysis was performed on the 47 jaundiced patients who received surgical resection with curative intent to identify the factors that influence the long-term survival of these patients (Table 6). The patients who underwent R0 resection survived longer than those who underwent R1 resection (p = 0.004). Multivariate analysis showed that R1 resection was the only independent predictor of poor prognosis.

Comparison between jaundiced patients with and without PBD (Table 7)

After PBD, direct bilirubin decreased from 216.5 ± 131.9 mol/L to 116.9 ± 66.3 mol/L (p < 0.001) with decreasing AST and ALT levels. Overall, the one- and three-year survival rates were 37.9% and 6.3% in the jaundiced patients with PBD (n = 19), respectively, and 38.5% and 8.7% in the jaundiced patients without PBD (n = 28), respectively. The primary endpoint of the three-year survival after surgery was not significantly different between the groups. The survival rates of the jaundiced patients with PBD were similar to those of the jaundiced patients without PBD (p = 0.968). The only patient who died of liver failure within the first 30 d postsurgery was a jaundiced GBC patient without PBD. The rate of postoperative intra-abdominal abscesses was not significantly different between the jaundiced patients with and without PBD (n = 4, 21.1% vs. n = 5, 17.9%, p = 0.787).

Implication of PBD before resection for advanced GBC with preoperative jaundice

Preoperative liver optimization has been a subject of debates for the last two decades, especially in PBD [12]. A prospective cohort study found that PBD significantly increases the rate of infectious complications [12]. Another study concluded that the routine use of PBD was not justified, considering that the mortality rate was not significantly different and the hepatic synthetic function recovery was identical to those of non-jaundiced patients [13]. To date, a few randomized controlled trials or meta-analyses have been conducted to systematically evaluate the value of PBD for the surgical resection of advanced GBC with preoperative jaundice. In the present study, the only patient who died of liver failure was a jaundiced GBC patient without PBD. Moreover, the postoperative mortality rates were not significantly different between the jaundiced patients with and without PBD.

Biliary obstruction is associated with renal failure, body fluid disturbances, and myocardial dysfunction. In our population, preoperative jaundice was a significant prognostic factor (p < 0.001) (Figure 4), which was consistent with the results our previous study [14]. We proposed that PBD via PTBD improves preoperative liver function; however, the effect of PBD on postoperative infection risk remains to be clarified [15–19]. In the present study, the decrease in the bilirubin level was statistically relevant (p < 0.001). However, this benefit was not associated with a longer survival time (p = 0.968).

Intra-abdominal abscesses often directly precede liver failure; thus, several researchers have shown that PTBD and increased intra-abdominal abscesses are significantly related [12, 16]. However, no significant differences in the rates of postoperative intra-abdominal abscesses were detected between jaundiced patients with and without PBD (21.1% vs. 17.9%, respectively, p = 0.787). Univariate analysis showed that PBD was not a risk factor for postoperative complications in jaundiced patients.

Liver surgery in jaundiced patients is supposed to have particular risks because of the hepatic and systemic changes caused by hyperbilirubinemia [15, 18]. Experimental studies on jaundiced animals have shown the benefits of biliary drainage, especially of internal biliary drainage with the restoration of biliary salt enterohepatic circulation [20]. PBD should increase cholestatic liver tolerance to ischemia [21] and reduce blood loss [22]. However, such benefits were not observed through the postoperative mortality and morbidity rates in jaundiced patients, which was not consistent with the results of previous studies [12, 23]. The lack of postoperative liver failure in the present study may be attributed to the low frequency of additional major surgical procedures, such as major hepatectomy and pancreaticoduodenectomy.

Preoperative jaundice as an indicator of poor prognosis and high postoperative morbidity but not a surgical contraindication

Jaundice is an indicator of advanced GBC with poor prognosis [1–4]. In the present study, the jaundiced patients had longer postoperative hospital stay and lower five-year survival rates than the non-jaundiced patients (6.0% and 36.0%, respectively). However, multivariate analysis revealed that preoperative jaundice was not a significant risk factor of poor outcome. Therefore, we have not solely considered jaundice to be a surgical contraindication. Several jaundiced patients had increased survival rates following resection. However, in the present study, preoperative jaundice was the only independent predictor of postoperative morbidity in GBC patients. Clinicians should be aware of this finding.

Several recent articles have been published with encouraging results. Agarwal et al. [14] reported that the presence of obstructive jaundice caused by GBC is neither a sole indicator of advanced disease nor a surgical contraindication. D’Angelica et al. [4] reported a five-year survival rate of 20% and a median survival time of 19 months after resection of GBC with clinical common bile duct involvement. They concluded that advanced GBC patients with preoperative jaundice and/or pEBI are candidates for resection when distant metastases are absent and R0 resection is achievable. This finding is consistent with the results of Hideki Nishio [9] and our previous report [24]. In the present study, the non-jaundiced patients, even those with locally advanced (T3/T4) GBC, exhibited good prognosis. Our data are consistent with the findings of Nishio et al. [9] and Agarwal et al. [14]. Multivariate analysis showed that non-curative resection was the only independent predictor of poor prognosis in jaundiced GBC patients. Therefore, even advanced GBC patients with preoperative jaundice should be resected whenever R0 resection is achievable.

In the present study, multivariate analysis of all 192 GBC patients revealed that lymph node metastasis and gallbladder neck tumors were independent prognostic factors. Therefore, an extended dissection of the regional lymph nodes must be routinely performed to achieve good prognosis. However, no specific patterns between recurrence incidence and gallbladder neck tumors were found. In addition, the mechanism by which gallbladder neck tumors contribute to GBC progression remains unknown. One possible explanation is that the gallbladder neck is in an anatomically "busy" area because of involved adjoining bile duct, portal vein, liver, duodenum, and colon, making surgical resection and radiotherapy difficult [24, 25]. Gallbladder neck tumors significantly increase surgical difficulty and reduce the opportunities for radical resection. Meanwhile, a relatively small tumor in the gallbladder neck infiltrates the hepatic hilum, thus causing obstructive jaundice [26]. In conclusion, advanced GBC with extrahepatic bile duct invasion and/or jaundice is a candidate for resection when R0 resection is achievable. However, the radical resection of advanced gallbladder neck cancer is still challenging with high postoperative morbidity and poor prognosis (Figure 5).

Extended resection for advanced GBC

Treatment of advanced GBC requires combined resection because it usually extends into adjacent organs, such as the liver, transverse colon, duodenum, extrahepatic bile duct, HA, and PV. For advanced GBC, extended resections that include extended hepatectomy, pancreatoduodenectomy, extended lymph node dissection, and combined vascular resection have been recently performed [27–29]. Extended resections have a high surgical mortality rate, and patients who have undergone such surgery remain to have poor prognosis [30–32]. The benefit of radical resection is apparently limited in advanced GBC patients. Hence, identifying patients who will potentially benefit from aggressive resection is important. Understanding the limits of the current surgical therapies may save patients with aggressive tumors from high morbidity and mortality rates related to surgeries that are unlikely to offer any benefit. It also allows patients and physicians to focus on palliation to improve the patients’ quality of life.

CRAO other than the liver and extrahepatic bile duct was an independent predictor of poor prognosis (p = 0.041). No three-year survivors were found among the patients undergoing CRAO, although CRAO was not a significant risk factor for poor outcome in multivariate analysis (Table 5). Therefore, aggressive resection should be carefully planned in patients who require CRAO. Our data are also consistent with the findings of Nishio [9].

Considering the high risk of death and poor long-term survival, we found no definitive benefit from the extended resections for advanced GBC [30]. A parenchymal preservation strategy has been investigated on the basis of the results published by Agarwal et al. [14]. A similar strategy for minor resection has been recently recommended by Regimbeau [8] for jaundiced patients with GBC. In the present study, we retained the most normal liver parenchyma to reduce the incidence of postoperative complications from R0 resection. We also obtained a comparable five-year survival rate and relatively low mortality and morbidity rates. Therefore, we recommend minor resection when a negative margin is ensured during operation.

This study has some limitations. Despite the large number of patients included in the present study, the patients were a highly selected cohort and represent only a fraction of the total number of jaundiced GBC patients diagnosed during the study period. Additional investigations with a large patient population in a multicenter study are needed before definitive conclusions can be drawn. Further studies must be conducted to clarify the resection contraindications in advanced GBC patients.

Ethical standards

All studies were approved by the Committee on Ethics of Second Military Medical University.