Indications for resection of recurrent lesions in patients with distal cholangiocarcinoma based on prognostic factors: a single-institute retrospective study and brief literature review

We enrolled 115 patients who underwent surgical resection for DCC at Tokai University Hospital between January 2000 and December 2018. The standard treatment for DCC is pancreatoduodenectomy (PD); however, bile duct resection (BDR) was chosen at the discretion of the attending physician for older patients or those for whom there were concerns regarding their ability to tolerate PD, and when the main lesion was in the middle bile duct. Staging of the primary DCC was performed according to the Union for International Cancer Control Tumor Node Metastasis (UICC TNM) classification, 7th edition. Postoperative complications were evaluated by the Clavien–Dindo classification. One surgery-related death and 13 deaths from other diseases were excluded, and 101 patients were finally analysed. Extensive cholangiocarcinoma treated with hepatectomy with concomitant PD was not considered in this study. The present study was approved by the Institutional Ethical Board of Tokai University Hospital.

All patients received routine postoperative surveillance. Patients underwent tumour marker measurements every three months. Chest and abdominal computed tomography (CT) were performed every 3 months for the first 3 years, every 6 months for the following 2 years, and annually thereafter. If necessary, magnetic resonance imaging or positron emission tomography CT was added to diagnose recurrence. Recurrence was diagnosed based on the radiological or pathological findings, and patients with only elevated tumour markers were not considered to have recurrence. Since there is currently no evidence for postoperative adjuvant chemotherapy, the decision to introduce adjuvant chemotherapy was made at the discretion of the attending physician.

In our department, the following criteria are defined as indications for resection of recurrent lesions: (1) the patient’s general condition is good; (2) the lesion is a single lesion, or two or more lesions are localised and curatively resectable; and (3) the time to recurrence is more than one year. Patients who do not meet these conditions may be treated with systemic chemotherapy or palliative care, depending on their general condition.

All statistical analyses were performed using a standard statistical programme (SPSS software for windows, version 26.0; Chicago, IL, USA). Chi-square tests were used to analyse categorical variables and Mann–Whitney U tests were used to analyse continuous variables. Overall Survival (OS) and recurrence-free survival (RFS) were analysed using the Kaplan–Meier method, and statistical significance was evaluated by the log-rank test. Univariate and multivariate Cox proportional hazard regression analyses were performed on variables to identify prognostic factors for primary DCC. Multivariate analyses were performed on variables with p < 0.05 from the univariate analyses. A p-value < 0.05 was considered statistically significant.

This study included 101 resected cases of DCC; 84 males and 17 females, with a median age of 72 years (16–86 years). The median observation period was 48 months (2–171 months). Baseline characteristics and clinicopathological features of these 101 patients at the time of surgical resection of primary DCC are shown in Table 1.

PD and extrahepatic BDR were performed in 94 (93.1%) and seven cases (6.9%), respectively. Five patients (5.0%) underwent portal vein resection and reconstruction. The median operative time was 298 min (183–592), and the median blood loss was 761 mL (114–6814 mL). Blood transfusion was performed in 24 patients (23.8%), and Clavien–Dindo classification of 3a or higher complications were observed in 65 patients (64.4%), which were attributed to pancreatic fistula grade B. Recent reports have reported an increased incidence of pancreatic fistula above grade B after PD for DCC ranging from 26.3% to 56.0% [14‐16]. The high incidence of complications has been attributed to this; fortunately, there were no deaths due to haemorrhage or sepsis. One factor that contributed to this may be the high number of young, relatively inexperienced surgeons who were in training and performed PD in our department.

The median length of stay was 30 days (9–100). There were no in-hospital deaths, but one case of early recurrence resulted in death within 90 days. This patient demonstrated an early rise in tumour markers and abdominal contrast-enhanced CT showed multiple liver metastases that had not been detected preoperatively. It was a case of very early recurrence. The median tumour diameter was 24 mm (1–90). Tumour tissues were well differentiated adenocarcinoma, moderately differentiated, poorly differentiated, and adenosquamous carcinoma in 59 (58.4%), 28 (27.7%), 3 (3.0%), and 1 case (1.0%), respectively. Lymphovessel, venous, and perineural invasions were observed in 73 patients (72.3%), 61 (60.4%), and 80 (79.2%) patients, respectively. Pancreatic and portal vein invasions were observed in 51(50.5%) and four (4.0%) cases, respectively. Lymph node metastasis was observed in 32 patients (31.7%), R0 resection was performed in 75 patients (74.3%), and among the R1 resections, seven patients (26.9%) had carcinoma in situ (CIS). In these cases, additional resection of the bile duct stump on the liver side was performed due to findings of invasive carcinoma or CIS in the bile duct stump at the time of initial resection or additional resection. However, these are cases where the marginal margins for additional resection in the extrahepatic bile duct were still CIS-positive. All patients were ≥ 75 years old, and additional hepatectomy was determined to be difficult in terms of surgical tolerance. No additional hepatectomy was performed, as several reports have shown that positive CIS at the bile duct stump in cholangiocarcinoma has a relatively favourable long-term prognosis [17, 18]. Postoperative adjuvant chemotherapy was administered in 34 patients (33.7%). The regimens were as follows: Tegafur/Uracil in two patients (5.9%), gemcitabine in five patients (14.7%), S-1 in 14 patients (41.2%), and gemcitabine combined with S-1 in 13 patients (68.2%).

During the observation period, recurrence occurred in 52 of the 101 patients (51.5%). The first sites of recurrence were local, liver metastasis, peritoneal dissemination or abdominal wall recurrence, lung, para-aortic lymph node, and combined recurrence in 18 (34.6%), 15 (28.8%), 5 (9.6%), 4 (7.7%), 2 (3.8%), and 8 (15.4%) patients, respectively (Fig. 1). The median RFS was 47.0 months (21.9–72.1), and the 1, 3, and 5-year RFS rates were 85.1%, 53.1%, and 45.3%, respectively (Fig. 2a). The median OS was 83.0 months (56.3–109.7), with 1, 3, and 5-year survival rates of 93.1%, 72.7%, and 54.4%, respectively (Fig. 2b). There was no significant difference in the OS by recurrence sites (p = 0.237).

A comparison of clinicopathological factors between R0 and R1 cases indicated that the tumour diameter was significantly larger in R1 cases (p = 0.001). In addition, the gross type was flat or nodular (p = 0.034), positive pancreatic invasion (p = 0.030), positive portal vein invasion (p = 0.004), positive perineural invasion (p = 0.009), pathological T3 (p < 0.001) were significantly more common in R1 cases. In summary, R1 cases tended to have higher oncological grades. A comparison of recurrence sites revealed that local recurrence was significantly more common in R1 cases (p < 0.001); no significant differences were found for other sites (Table 2). The median RFS did not reach the median for R0 cases, whereas it was 16 months (9.9–22.1) for R1 cases, with R0 cases having a significantly better prognosis (p < 0.001). Similarly, the median OS of R0 patients was 91 months (81.6–100.4) compared with 33 months (8.3–57.7) for R1 patients, with R0 patients having a significantly better prognosis (p < 0.001).

In the 52 patients who relapsed, 7 (13.5%), 28 (53.8%), and 17 (32.7%) underwent resection of recurrent lesions, chemotherapy, and transferred to palliative care, respectively. First-line chemotherapy was administered to 15, 6, 6, and one patient with gemcitabine plus cisplatin, gemcitabine, S-1, and gemcitabine plus S-1, respectively. Resection of recurrent lesions included liver metastasis, abdominal wall recurrences, and PD for local recurrence after extrahepatic BDR in five, one, and one case, respectively. The median age was 75 years (63–79), and all patients were male. All patients had pathological T2 or less, six of the seven cases were negative for lymph node metastasis and R0 resection was performed. At the discretion of the attending physician, resection was performed in only one case of recurrence within one year. Only two patients had resection of second recurrent lesions.

There were no cases of recurrent lesion resections requiring perioperative blood transfusion, and there were no complications requiring therapeutic intervention of the Clavien–Dindo classification 3a or higher (Table 3). Median RFS was 19.0 months (16.4–21.6 months), and the median OS was 83.0 months (0.0–185.6 months). The median OS of patients without surgery for recurrent lesions was 34 months (19.0–49.0 months), and the prognosis of patients with surgery was significantly better (p = 0.022) (Fig. 3a). Similarly, the median OS after recurrence was 30 months (19.7–40.3 months) in the resection group, and 10 months (6.0–14.0 months) in the non-resection group, with the resection group having a significantly better prognosis (p = 0.005). A patient who underwent additional PD for local recurrence after extrahepatic BDR died of early local recurrence. In contrast, four of the seven patients achieved a five-year survival. In patients with recurrent liver metastases, the prognosis of patients with resected liver metastases was significantly better than that of patients without resected liver metastases, with an OS of 98 months (0.0–199.4 months) and 26 months (23.1–28.9 months), respectively (p = 0.038) (Fig. 3b).

Analysis of prognostic factors in the 101 resected cases of DCC was performed. In univariate analysis, operative time, blood transfusion, tumour size, tumour gross type, portal vein invasion, lymphovessel invasion, venous invasion, perineural invasion, lymph node metastasis, residual tumour status, UICC-T factor, tumour differentiation, and recurrence within one year were significantly associated with prognosis. Multivariate analysis of the prognostic factors showed that positive portal vein invasion (hazard ratio [HR]: 6.44, 2.0–21.1, p = 0.002), positive lymph node metastasis (HR: 2.75, 1.4–5.2, p < 0.001), and recurrence within one year (HR: 30.77, 11.5–82.4, p < 0.001) were independent poor prognostic factors.

Analysis of prognostic factors in the 52 recurrent cases of DCC was performed. In univariate analysis, tumour size, tumour gross type, UICC-T factor, and recurrence within one year were significantly associated with prognosis. Multivariate analysis of prognostic factors showed that recurrence within one year (HR: 16.4, 6.5–41.4, p < 0.001) was an independent, poor prognostic factor. Resection of recurrent lesions improved prognosis (HR: 0.29, 0.1–0.9, p = 0.025). In recurrent cases, lymph node metastasis and portal vein invasion were not independent poor prognostic factors (Table 4).