AIM: To analyze whether pancreaticoduodenectomy with simultaneous resection of tumor-involved vessels is a safe approach with acceptable patient survival.

METHODS: Between January 2001 and March 2012, 136 patients received pancreaticoduodenectomy for adenocarcinoma at our hospital. Seventy-eight patients diagnosed with pancreatic head carcinoma were included in this study. Among them, 46 patients received standard pancreaticoduodenectomy (group 1) and 32 patients received pancreaticoduodenectomy with simultaneous resection of the portal vein or the superior mesenteric vein or artery (group 2) followed by reconstruction. The immediate surgical outcomes and survivals were compared between the groups. Fifty-five patients with unresectable adenocarcinoma of the pancreas without liver metastasis who received only bypass operations (group 3) were selected for additional survival comparison.

RESULTS: The median ages of patients were 67 years (range: 37-82 years) in group 1, and 63 years (range: 35-86 years) in group 2. All group 2 patients had resection of the portal vein or the superior mesenteric vein and three patients had resection of the superior mesenteric artery. The pancreatic fistula formation rate was 21.7% (10/46) in group 1 and 15.6% (5/32) in group 2 (P = 0.662). Two hospital deaths (4.3%) occurred in group 1 and one hospital death (3.1%) occurred in group 2 (P = 0.641). The one-year, three-year and five-year overall survival rates in group 1 were 71.1%, 23.6% and 13.5%, respectively. The corresponding rates in group 2 were 70.6%, 33.3% and 22.2% (P = 0.815). The one-year survival rate in group 3 was 13.8%. Pancreaticoduodenectomy with simultaneous vascular resection was safe for pancreatic head adenocarcinoma.

CONCLUSION: The short-term and survival outcomes with simultaneous resection were not compromised when compared with that of standard pancreaticoduodenectomy.

Carcinoma of the pancreas is one of the leading causes of death, with 43140 new cases reported in the United States in 2010, which will lead to an estimated 36800 deaths at a five-year survival of 6%[1]. Amongst different treatment options, surgical resection offers the best survival outcome to patients with carcinoma of the head of the pancreas[2]. With the advancement in technology and experience sharing, the hospital mortality and morbidity for pancreaticoduodenectomy, also known as the Whipple operation, have improved when compared with the first report in 1935[3]. However, the surgery remains a challenging operation, with hospital mortality rates ranging from 1% to 6% even at experienced centers[4,5]. The issue is even more complicated if the tumor involves major vessels around the pancreatic region. The definition of borderline resectability is controversial. Although many centers have advocated resection of the tumor together with the superior mesenteric vein (SMV) or the portal vein (PV), many other centers simply do not consider operation for this group of patients after balancing the risk of surgery and predicted survival outcomes[6,7].

The aim of this study was to compare standard pancreaticoduodenectomy and pancreaticoduodenectomy with simultaneous vascular resection with or without vascular reconstruction in terms of survival outcomes in patients who had adenocarcinoma of the pancreas with borderline resectability.

The medians for volume of blood loss (1200 mL vs 800 mL; P < 0.05) and operation time (715 min vs 580 min; P < 0.05) were significantly greater in group 2 compared to group 1 (Table 3). Both groups had an intensive care unit stay of 2 d and a median hospital stay of 17 d, with no differences observed in postoperative complications.

The median tumor size for both groups was 3 cm, and the majority of patients had fewer than five lymph node metastases (Table 4). There were no differences between the groups regarding the method of pancreaticojejunostomy, type of resection, or disease stage as classified the American Joint Committee on Cancer (AJCC) classification (7^th edition)[14].

The one-, three-, and five-year survival rates in group 1 were 71.1%, 23.6% and 13.5%, respectively (Figure 2A). These were not different from group 2, who showed corresponding rates of 70.6%, 33.3% and 22.2%. The one-, three-, and five-year disease-free survival rates in group 1 were 48.7%, 15.7% and 15.7%, respectively (Figure 2B), which corresponded to 40.3%, 20.2% and 13.4% disease-free survivals in group 2.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Comparison of survival rates. A: Overall and B: Disease-free survival curves for patients in group 1 and group 2; C: Overall survival curves for patients in group 2 and group 3; D: Stage-specific survival curves for group 1 and group 2 patients; Group 1a and 2A: Stage I and IIA disease; Group 1b and 2b: Stage IIB and III disease.

The one-year survival of patients in group 3 (stage IIB/III) who had only undergone bypass surgery was 13.8%, which was significantly lower than in group 2 (P < 0.05) (Figure 2C). These patients were either considered physically unfit for pancreaticoduodenectomy or found to have an SMA encasement larger than 2 cm during laparotomy. Patients with liver metastasis and distant metastasis were not included in this group.

Patients were classified into early and advanced groups for survival comparison according to the presence of lymph node metastasis. The early group was comprised of patients with stage IA, IB or IIA disease. The advanced group was comprised of patients with disease stage IIB or III. The median survival durations for group 1 and group 2 patients with early disease were similar at 28.8 and 27.5 mo, respectively, and 15.8 and 17.1 mo for advanced disease stages (Figure 2D).

Sixteen factors that might affect patient survival after pancreaticoduodenectomy were identified (Table 5). Multivariate analysis showed that only lymph node metastasis was significant for poorer survival (P = 0.021), whereas a univariate analysis identified AJCC disease stage as the only significant factor.

Pancreaticoduodenectomy is a technically challenging procedure, and the rate of morbidities (including pancreatic fistula) ranges between 15% and 50%[15,16]. A gradual decrease in mortality has been observed in recent years, though the patient age and the incidence of comorbid illness have increased. It is generally agreed that a hospital mortality rate of 3%-5% is acceptable for pancreaticoduodenectomy without vascular resection at high-volume centers[4]. However, it is believed that the risk of this operation is high if simultaneous vascular resection with reconstruction is required. Many surgeons have taken a more conservative approach when the tumor is considered barely resectable[17,18].

Tumor invasion of the PV or SMV is considered a sign of advanced disease, which is likely to result in poor surgical outcomes. Thus, many centers adopt more conservative approaches for patients with vascular invasion[19]. In this study, we found that tumor invasion or dense adhesion to the PV or SMV did not correlate with the stage of disease. The median tumor size was 3 cm in both groups, but tumors as small as 1.6 cm could have invaded the PV or SMV, necessitating vascular resection. It is tumor location rather than tumor size that affects surgical planning. Fuhrman et al[20] described a similar observation in 23 patients whose tumors with a median size of 3 cm led to major vessel involvement.

Due to the retrospective design of this study, we did not present the degree of invasion of the PV or SMV found by histopathologic examinations. Since histologic information can only be obtained after surgery, an operating surgeon has to judge during laparotomy whether there is genuine vascular invasion and whether simultaneous vascular resection is required. At centers experienced in vascular reconstruction, simultaneous resection for suspected venous invasion should be performed. The AJCC has not considered tumor invasion of the PV or SMV a factor affecting tumor staging, and therefore a tumor with PV or SMV involvement is classified stage IIA if there is no lymph node metastasis. In accordance with this, patients with tumor invasion of the PV or SMV are still considered to have a relatively early-stage cancer. This might be quite contradictory to many clinicians’ concepts.

The definition of borderline resectability for pancreatic cancer remains controversial. Varadhachary et al[21] proposed three categories of patients with tumors with borderline resectability: types A, B and C. Type-A patients have tumors with one or more of the following three findings on CT images: (1) tumor abutment (≤ 180° of the circumference of the vessel) of the SMA or celiac axis; (2) tumor abutment or encasement (> 180° of the circumference of the vessel) of a short segment of the hepatic artery, typically at the origin of the gastroduodenal artery; and (3) tumor-related occlusion of a short segment of the SMV, PV, or SMV-PV confluence that is amenable to vascular resection and reconstruction because of a patent SMV and PV below and above the area of occlusion. Type-B patients have tumors with extrapancreatic metastasis. Type-C patients are patients who have marginal physical fitness for major operations.

In fact, many patients with adenocarcinoma of the pancreas can have a relatively early cancer stage even if they are classified as a type-A patient. The decision on treatment strategy for these patients depends on the risk and benefit of surgery and whether or not there is a good alternative treatment. Neoadjuvant chemotherapy has been suggested to increase the resection rate. Chemotherapy or chemoradiation have a partial tumor response rate of 56%[22]. Although it might be effective for some patients, subjecting every patient with SMV or PV involvement to neoadjuvant therapy without considering upfront surgery would allow progression of cancer in 40% of patients who would not respond well.

This study shows that pancreaticoduodenectomy with vascular resection can be performed safely at centers with experience and expertise. The complication and pancreatic fistula rates in group 2 were not inferior to those in group 1. No SMV or PV thrombosis was found in the study. The experience and techniques in vessel reconstruction we have learned from our liver transplant program can be transferred to many complicated hepatobiliary and pancreatic surgeries[10,23,24]. Techniques used in reconstruction of the PV, SMV or SMA are identical to those used in liver transplantation. Most of the time, PV or SMV reconstruction is possible without generating tension on the venous anastomosis after a Cattell-Braasch maneuver[25]. When simultaneous resection of a PV longer than 5 cm is required during pancreaticoduodenectomy, cadaveric vein grafts available from the liver transplant program give extra flexibility for the vascular reconstruction.

Contrast CT is one of the most sensitive ways for preoperative staging of pancreatic cancer[26,27]. It has been suggested that the morphologic features of portal vein invasion could predict the survival outcomes of patients. Patients with extensive tumor involvement of the vessel would have poorer survival[27,28]. Likewise, possible regional lymph node metastases can now be revealed by fine-cut CT scans. Regional lymph node metastases are associated with poor patient survival. However, these so-called poor prognostic indicators should only be regarded as a prognostic suggestion before surgery, and should not become an absolute contraindication to surgery, as surgical resection provides the best survival outcomes for patients with barely resectable diseases. Pancreatic cancer with arterial invasion is also associated with poor patient survival. In a previous study, patients who had pancreaticoduodenectomy with simultaneous arterial resection had a median overall survival duration of only 15.8 mo[29].

In our present study, patients having only bypass surgery with palliative intention (group 3) were included for comparison. Many of these patients were not subjected to pancreaticoduodenectomy because they were not physically fit for major surgery. In terms of survival, these patients fared far worse than group 2 patients. Approximately 23% of the patients in the study had R1 resection as shown by final pathologic examination. Many of these patients had posterior margin involvement. No difference in the distribution of margin involvement was found between group 1 and group 2. As there is no association between SMV or PV invasion and R1 resection, and the degree of posterior margin involvement cannot be known before laparotomy, patients should not be denied pancreaticoduodenectomy on the basis of CT images of vascular invasion.

Adenocarcinoma of the pancreas is a cancer with poor patient survival. Surgical resection provides the best chance of cure. Pancreaticoduodenectomy with simultaneous vascular resection is a safe and effective treatment option. The postoperative morbidity and pancreatic fistula rates are not inferior at centers with expertise. In patients who suffer from adenocarcinoma of the pancreas with portal venous invasion, survival after this complicated procedure is not compromised when compared with that after standard pancreaticoduodenectomy.