From August 2005 to December 2019, 297 patients with pancreatobiliary malignancy underwent surgery with a curative intent at our hospital. Of 297 patients, 194 with histologically proven adenocarcinoma were included in this retrospective cohort study. Their medical records, computed tomography (CT) findings, surgical procedures, pathological findings, adjuvant chemotherapy data, and survival outcomes were retrospectively reviewed. We classified cancers of pancreatobiliary origin as ampulla of Vater cancer, gallbladder cancer, pancreatic cancer, and cholangiocarcinoma (e.g., Klatskin tumor, extrahepatic and intrahepatic cholangiocarcinoma).

Regarding pathological findings, we assessed the pathological tumor-node-metastasis (pTNM)-stage, presence of microscopically remnant disease, perineural invasion, total count of resected lymph nodes, and total count of positive lymph nodes. The pT stages of T3 and T4 were classified as high and those of T1 and T2 as low. The TNM stage was evaluated based on the American Joint Committee on Cancer (AJCC) Cancer Staging Manual, 7th edition [10].

Recurrence was defined as confirmation of the presence of a lesion by radiology or biopsy. Local recurrence was defined as reappearance of a tumorous lesion at the anastomosis site or around the PAFC based on imaging findings.

PAFC was defined as a collection of fluid within a diameter ≥ 3 cm, as demonstrated by CT [5,11]. The presence or absence of PAFC on CT was reviewed by radiologists. The duration index date was defined according to the postoperative day. At our hospital, CT is performed routinely to assess postoperative complications within 1 week after surgery. A second CT examination is performed at 4 weeks postoperatively to determine whether the postoperative complications, including PAFC, have improved. Based on this CT schedule, longstanding PAFC was defined as PAFC that did not regress spontaneously and was detected by CT or ultrasound for > 4 weeks after surgery. We excluded patients who did not undergo CT examinations at 4 weeks postoperatively. Patients with preoperative ascites and those receiving ambulatory peritoneal dialysis were excluded due to the difficulty of identifying PAFC in such patients.

All follow-up data, including the date of last follow-up, vital status, cancer recurrence, and survival, were obtained by contacting doctors and hospitals and from registry medical records. Survival duration was measured from the date of surgery until death or the date of the last hospital visit. Patients lost to follow-up were also excluded.

The recurrence of pancreatobiliary adenocarcinoma was the primary outcome measure. Categorical variables were expressed as frequencies and percentages, and continuous variables were expressed as mean ± standard deviation (or ranges). The variables were compared using the chi-square test. Variables with p values < 0.2 were entered into a logistic regression model for multivariate analyses. Cancer-specific survival (CSS) was considered as a secondary outcome and was calculated using the Kaplan-Meier method. Variables with p values < 0.2 were entered into a Cox proportional-hazards regression model for multivariate analyses. Patients who died due to surgical complications or causes other than cancer were excluded.

Statistical analysis was conducted using SPSS statistical software version 24.0 (IBM Co., Armonk, NY, USA). p values < 0.05 were regarded as statistically significant.

This study was performed according to the principles of the Declaration of Helsinki. The Institutional Review Board of Konkuk University Medical Center approved this study (approval number: KUH1010922). The requirement of informed patient consent was waived by the Institutional Review Board due to the retrospective nature of the study.

In total, 297 patients with pancreatobiliary duct cancer underwent surgery with curative intent; 194 of these were eligible for the present study. Of these, 103 patients were excluded for the following reasons: 42 patients were lost to follow-up, 26 patients were diagnosed with conditions other than adenocarcinoma (e.g., neuroendocrine tumor, pancreatic serous cystadenoma, solitary pseudopapillary tumor, intraductal papillary mucinous neoplasms, and benign pathology), 30 patients did not undergo CT examinations at 4 weeks after surgery, three patients were diagnosed with one or more other types of cancer, one patient died due to surgical complication, and one patient was receiving peritoneal dialysis (Fig. 1).

The baseline clinical data are summarized in Table 1. The mean age of the patients was 68 ± 9.7 years. One hundred and four (53.6%) patients were men and 90 (46.4%) were women. Sixty-six (34.0%) patients had hypertension, 22 (11.3%) had diabetes mellitus, and 52 (26.8%) had dyslipidemia. The mean follow-up period was 37 ± 28.9 months. The most common tumor type was cholangiocarcinoma (97 patients, 50.0%). Fifty-seven patients had distal cholangiocarcinoma, 16 had Klatskin tumor, and 24 had intrahepatic cholangiocarcinoma. Sixty (30.9%) patients had pancreatic cancer and 37 (19.1%) had ampulla of Vater cancer. The resection margin status was positive in 21 (10.8%) patients. PAFC was detected by CT in 165 (85.1%) patients and was longstanding (> 4 weeks) in 74 (38.1%) of these patients. Of the 74 patients with longstanding PAFC, the fluid collection improved with medical treatment in only 22 patients. In the remaining 52 patients, PAFC did not disappear during the follow-up period; however, no surgical intervention was attempted because complicated fluid collections did not occur in this group. Cancer recurred during the follow-up period in 114 (58.8%) patients: local recurrence in 92 (47.4%) patients and metastatic recurrence in 22 (11.3%) patients. Seventy-seven (39.7%) patients died due to cancer recurrence. The surgical approach for resection of pancreatobiliary cancer and the prevalence of longstanding PAFC are summarized in Table 2.

In univariate analyses, the recurrence rates were 58.8% (n = 97) in patients with PAFC and 58.6% (n = 17) in patients without PAFC (p = 0.987). The recurrence rate was significantly higher in patients with longstanding PAFC (68.9%, n = 51; p = 0.025) than in patients with non-longstanding PAFC (50.5%, n = 46; p = 0.025). Fig. 2 shows a local recurrence in a patient with longstanding PAFC. Recurrence was also significantly associated with high T stage (T3, T4; p = 0.040), lymph node involvement (p < 0.001), perineural invasion (p < 0.006), and non-receipt of adjuvant chemotherapy (p = 0.025). Age (> 70 years), sex, resection margin status, and presence of PAFC were not associated significantly with recurrence (Table 3).

In the multivariate analysis, the recurrence rate was associated with lymph node involvement (p = 0.003) and longstanding PAFC (p = 0.020). No other variable was associated with the recurrence rates (Table 4).

The presence of PAFC was not significantly related to recurrence of any subtype of pancreatobiliary adenocarcinoma. However, longstanding PAFC was significantly related to recurrence of pancreatic adenocarcinoma (p = 0.016) (Table 5) in univariate analyses.

In univariate analyses, poor CSS was significantly associated with a high T stage (T3, T4; p = 0.001), lymph node involvement (p < 0.001), perineural invasion (p = 0.004), and non-receipt of adjuvant chemotherapy (p = 0.002). Age (> 70 years), sex, and resection margin status were not significantly associated with poor CSS (Table 6).

In the multivariate analysis, poor CSS was associated with a high T stage (T3, T4; p = 0.042) and lymph node involvement only (p < 0.001) (Table 7). However, in both univariate and multivariate analyses of patients with pancreatic adenocarcinoma, no variable was associated with poor CSS.