Between January 2003 and December 2009, 163 patients underwent surgery - PD with curative intent - for bile duct cancer at Keimyung University Dong San Medical Center. For 55 of them, PD including pylorus-preserving PD (PPPD) was performed for treatment of distal CBD cancer, and these patients were enrolled in our study. These patients were divided into two groups according to their age: a younger group (age less than 70 years old) and an older group (age 70 or greater).

There were no absolute criteria for performing preoperative biliary drainage. In general, if the patients showed jaundice and a bilirubin level higher than 7-10 mg/dl, or if it was necessary to do a biopsy, biliary drainage was applied. Biliary drainage was done by endoscopic retrograde biliary drainage, endoscopic nasobiliary drainage or percutaneous transhepatic biliary drainage. Preoperative diagnosis was done based on computed tomography and magnetic resonance cholangiopancreatography or cholangiography. PD was performed when the level of bilirubin was decreased down to 5 mg/dl.

The PD procedure consisted of standard PD or PPPD, including regional lymph node dissection around the porta hepatis, retropancreatic and celiac trunk area and paraaortic lymph node if necessary. In most cases, PPPD was performed and when the tumor was suspected to be advanced locally, especially in the first portion of the duodenum or distal stomach, PD was performed. The reconstruction of the pancreas was done by the Dunkin method or by a duct-to-mucosa technique. In the Dunkin method, a polyethylene tube was inserted into the pancreatic duct and brought out externally via a transhepatic route in order to drain of the pancreatic juice externally for 2-3 weeks.16 In the duct-to-mucosa method, an internal stent was inserted into the pancreatic duct when the diameter of the pancreatic duct was less than 4 mm, but not if the diameter of the pancreatic duct was larger than 4 mm, stent.

Collected data included demographics (sex, tumor marker, tumor size, preoperative bilirubin level, the stage of the cancer (AJCC 7^th), tumor differentiation, perineural invasion, lymphovascular invasion, and lymph node metastasis), surgical results (surgical procedure, duration of the operation, amount of bleeding, intraoperative transfusion and curability, morbidity) and hospital stay length. Outcomes of the two groups were compared for perioperative morbidity, mortality and survival rates. Postoperative surgical complications, including pancreatic fistula and delayed gastric emptying, were defined and graded, as described by the proposed Clavien classification system17: grade I as any deviation from the normal postoperative course without clinical significance, grade II as requiring pharmacological treatment, grade III as requiring surgical, endoscopic or radiological intervention, grade IV as life-threatening complications requiring intermediate or intensive care unit management, and grade V as death of a patient. Furthermore, pancreatic fistula was defined according to the International Study Group of Pancreatic Fistula (ISGPF) classification scheme (grade A, B, C)18 and the grade of delayed gastric empting was defined according to the International Study Group of Pancreatic Surgery (ISGPS)19 classification scheme (grades A, B, C). Surgical mortality was defined as death noted within 30 days postoperatively.

All patients were monitored at 3-6 month intervals by measuring levels of CA 19-9, CEA and performing a CT scan annually for the first 2 years and then at 6-12 months interval.

Numeric variables were expressed by mean and standard deviation. Patients' demographics and surgical results were evaluated using the chi-square test and Mann-Whitney test, using SPSS for Windows version 18.0 (SPSS Inc., Chicago, Illinois). Survival analysis was done using the Kaplan-Meier method. Differences in survival between groups were compared using the log-rank test. Significance was accepted when the probability was less than 0.05.

Characteristic of the 55 patients are summarized in Table 1. Of the 55 patients having curative resections, 19(34.5%) were in the older group and the remaining 36 patients (65.5%) were in the younger group. The mean ages of the two groups of patients were 73.5±2.4 years and 60.5±6.4 years respectively. The older patients had significantly more comorbid diseases such as diabetes and cardiovascular disease. Demographic factors and clinicopathologic findings were similar in the two groups.

Postoperative short-term outcomes are shown in Table 2. There were no significant differences between the groups in terms of operation time, estimated amount of intraoperative bleeding and duration of postoperative hospital stays. There was no significant difference in the postoperative surgical complication rate (57.9% in Older group Vs 52.8% in Young group, p=0.717) and grade of complication according to the Clavien classification. One patient of the younger group was dead on the 3^rd postoperative day due to myocardial infarction. A higher level (more than 5 mg/dl) of preoperative initial bilirubin showed significant correlations with operative morbidity by univariate analysis and age was not an independent risk factor of operative morbidity (Table 3). However, no factors were significantly correlated with operative morbidity by multivariate analysis.

During a mean follow-up period of 34.4 months, the 5-year survival rate was analyzed in the two groups. Overall 5 year survival of the older and younger groups were 45.9% and 39.5% respectively (p=0.671) and disease free 5 year survival rates were 31.7% and 31.1% respectively (p=0.942). There were no significant differences between the two groups (Fig. 1).