This retrospective study was approved by the institutional review board of the second affiliated hospital of Guangzhou Medical University. Between January 2010 and September 2014, 242 patients with extrahepatic CCA were treated in our hospital. Seventy-six patients were included in this study using the following criteria and the detailed selection process is described in Figure 1.

1) Inclusion criteria: a) Patients with histologically or cytologically confirmed extrahepatic biliary adenocarcinoma but ineligible for curative surgery due to locally advanced or metastatic disease or unfit for/not willing to undergo a major operation; b) patients who successfully underwent SEMS placement with or without ERFA as initial treatment for obstructive jaundice by endoscopic retrograde cholangiopancreatography (ERCP) or percu- taneous transhepatic cholangiography (PTC) approaches if ERCP fails.

2) Exclusion criteria: a) Patients with obstructive jaundice caused by other malignant tumors (tumors of the ampulla of Vater, pancreatic tumors, hepatocellular carcinoma with endobiliary tumor thrombus, bile duct compressed by hilar lymph metastases, e.g.), or patients who had developed obstructive jaundice due to recurrent tumors after initial resection of the CCA; b) patients with Bismuth type II, III, or IV hilar CCA [21] [22] ;

c) patients who underwent unsuccessful SEMS placement with or without ERFA, who underwent other local ablative treatments (PDT) instead of ERFA, or who underwent placement of plastic stents instead of SEMS for obstructive jaundice.

Pre-treatment assessment of all patients included a complete history and physical examination for general health, laboratory tests and radiologic imaging. A dual-phase contrast-enhanced spiral computed tomography and/or magnetic resonance cholangiopancreatography were performed out on all patients to evaluate biliary tumors and possible vascular involvement. If needed, positron emission tomography with [(18)F]fluoro-2-deoxy-D-glucose (FDG-PET) was used to identify occult metastatic disease. All data were reviewed by a multidisciplinary treatment team of radiologists, surgeons, pathologist and oncologists to select treatment options. For those with unresectable extrahepatic CCA, the option of ERFA before SEMS placement and its potential benefits for prolonging stent patency were discussed with patients, who would make the ultimate decision. Informed consent was obtained from all patients.

All patients received intratracheal general anesthesia. Side viewing endoscopes TJF-160 (Olympus America, Center Valley, PA) were used for all endoscopical procedures. Patients underwent biliary sphincterotomy followed by cholangiogram to confirm stricture location. An endoscopical clamp or cytology brush was use to obtain biopsies for histological analysis. Percutaneous biopsy and cholangiography were performed under real-time US guidance (EUB-2000, HITACHI Medical Systems) for all percutaneous approaches.

For patients who received ERFA, the Habib EndoHPB catheter (EMcision UK, London, United Kingdom) was advanced to the stricture location, and radiofrequency energy was delivered by an RFA generator (1500 RF generator; RITA Medical Systems Inc, Fremont, CA) at 400 kHz at 10 W for 2 min, with a rest period of one min before moving the catheter. Repeat energy was applied to ensure the whole stricture length was covered. An uncovered (Cook Ireland ltd, Limerick, Ireland) or covered SEMS (WallFlex RX RMV, Boston Scientific Corporation, city, MA) was placed after ERFA (Figure 2).

All patients received computed tomography and/or MRI and laboratory blood tests about 30 days after the

procedure and bile duct stricture improvement was evaluated. Those with obvious bile duct decompression and persistent liver function improvement (total bilirubin and combined bilirubin decreasing) were considered to have full stricture improvement. Partial or no improvement was identified when radiologic images showed that a stricture remained in the previous tumor site and bilirubin did not normalize or bilirubin increased after decreased.

Adjuvant chemotherapy based on practice guidelines of the National Comprehension Cancer Network (NCCN) in hepatobiliary cancers was recommended for patients who recovered from the intervention [20] . All patients were regularly followed up every two months since the last intervention procedure until November 2014. When bile ducts occluded again, the multidisciplinary treatment team met to decide whether a repeat ERFA and/or SEMS placement or percutaneous biliary drainage was appropriate. Major complications were defined as procedure-related complications resulting in an unplanned increase in medical care, prolonged hospitalization, permanent adverse sequelae, or death [21] .

Demographic data for both groups was compared using the Student’s t test for continuous data and the Chi-squared test for categorical data. The median stent patency was compared by Mann-Whitney test. Overall survival curves from the date of ERFA and/or SEMS were calculated and constructed by the Kaplan-Meier method and compared using the log-rank test. The relative prognostic significance of the variables in predicting overall survival was assessed by multivariate Cox proportional hazards regression analysis. Statistical analyses were performed using SPSS 10.0 statistical software (SPSS Company, Chicago, IL). Continuous data are given as means ± S.D. All statistical tests were two-sided, and a significant difference was considered when P < 0.05.

Seventy-six patients with indication for SEMS and ERFA were included in this study, among which, 34 patients opted to receive ERFA followed by SEMS placement and 42 patients opted to receive SEMS only. There were no significant differences among the patients in demography (gender, age), tumor grade, and staging, patients’ general health conditions, and liver function parameters (Table 1). Most patients (72/76, 94.7%) had unresecta- ble extrahepatic CCA due to local advanced tumors (n = 65) or metastatic disease (n = 7). Two patients with po- tentially resectable extrahepatic CCA refused tumor resection, and 2 patients were considered unfit for a major operation (one had severe cirrhosis; the other had poor pulmonary function). Histological and/or cytological dia- gnosis of biliary adenocarcinoma was made in 66 patients by ERCP. For those who underwent a PTC approach, 6 patients also had hepatic metastatic tumors and the other 4 patients had Bismuth type I hilar CCA. Percuta- neous biopsy was successfully performed on hepatic metastatic tumors or hilar CCA. There was no difference in

^aAmerican Joint Committee on Cancer (AJCC) TNM Staging (7th edition 2010). 27; ^bPatients who declined to have major operation; ^cPatients who were unfit for major operation; ^dEndobiliary radiofrequency ablation; ^eSelf-expanding metal stent; ^fCholangiocarcinoma; ^gEndoscopic retrograde cholangiopancreatography; ^hPercutaneous transhepatic cholangiography.

follow-up duration between the two groups (ERFA + SEMS: 12.2 months ± 0.53, range, 6 - 19.5 months; SEMS only: 11.3 months ± 0.34, range, 7 - 16 months) (t-test, P = 0.078).

Thirty-day stent patency was obtained in all patients and the median stent patency in ERFA + SEMS group (9.5 months, range: 4.5 - 14.0 months) was significantly longer than that in the SEMS group (8.35 months, range: 4.9 - 11 months) (Mann-Whitney test, P = 0.024). During follow up, 4 patients in the ERFA + SEMS group received repeated ERFA, but only 2 patients completed the process proposed by Steel and co-workers, which required an ablation of 2 min at 10W [17] . The other two patients failed due to technical issues: a short-circuit produced in the primary stents during the early process, which led to the termination of the repeated ERFA. Four patients received a secondary SEMS placement within the primary stent successfully. Sixty-one patients received percutaneous biliary drainage when biliary obstruction relapsed during the followup. Twenty-three patients refused to have adjuvant therapy either due to costs or anxiety about adverse effects.

There was no treatment procedure-related mortality in the study subjects and no significant difference in treatment complications was observed between the two groups (Table 2). Twelve patients had moderate or severe post-procedure abdominal pain, which required oral or intravenous analgesics for relief of symptoms. Three patients developed cholangitis and one had pyogenic cholecystitis which required percutaneous transhepatic gallbladder drainage. All of these patients recovered fully after antibiotic treatment. Three patients were diagnosed with asymptomatic pancreatitis (serum amylase > 1000 IU/L) and recovered after conservative treatment.

Nine patients survived to the end of the follow-up period in November 2014. Of the patients who died, a detailed list of causes of death is summarized in Table 3. The major causes of death were disease progression (liver failure), bleeding, biliary infection, cachexia, and lung infection. The rates of occurrence of each of these showed no significant differences between the two both groups. None of the patients died from ERFA- or SEMS-related procedures.

Kaplan-Meier plots for ERFA + SEMS and SEMS groups associated with patient survival are shown in Figure 3. The log-rank test demonstrated that the overall survival in the ERFA + SEMS group (n = 34) was significantly longer than that in the SEMS group (n = 42) (P = 0.036), as shown in Figure 3(a). Further subgroup analysis excluded patients with metastatic disease (n = 7), or unfit (n = 2) or declined surgery (n = 2). The

^aEndobiliary radiofrequency ablation; ^bSelf-expanding metal stent.

^aEndobiliary radiofrequency ablation; ^bSelf-expanding metal stent.

log-rank test demonstrated that the overall survival for patients with local advanced extrahepatic CCA in the ERFA + SEMS group (n = 30) was significantly longer than in the SEMS group (n = 35) (P = 0.034), as shown in Figure 3(b). If patients with covered metal stents (n = 10) were excluded from the study, the overall survival in the ERFA + SEMS group was longer than in the SEMS group, but the difference was not significant (log-rank test, P = 0.084) as in Figure 3(c). Similarly, no significant differences were found between the two treatments with respect to unresectable Bismuth type I hilar CCA (log-rank test, P = 0.116) as in Figure 3(d); or unresectable distal CCA (log-rank test, P = 0.214) as in Figure 3(e).

Multivariate Cox proportional hazards regression analyses showed that ERFA, tumor AJCC stage, extrahepatic CCA type, serum albumin and adjuvant chemotherapy were prognostic factors for overall patient survival (Table 4).