Vascular Resection and Reconstruction for Pancreatic Malignancy: A Single Center Survival Study
- First Online:
- Cite this article as:
- Al-Haddad, M., Martin, J.K., Nguyen, J. et al. J Gastrointest Surg (2007) 11: 1168. doi:10.1007/s11605-007-0216-x
- 199 Views
Pancreatic cancer is one of the leading causes of cancer-related death in the USA. Recently, several centers have introduced portal and superior mesenteric vein resection and reconstruction during extended pancreatectomy, rendering the previously inoperable cases resectable.
The aim of this study is to confirm whether patients with locally advanced pancreatic cancer and mesenteric vascular invasion can be cured with extended pancreatectomy with vascular reconstruction (VR) and to compare their survival to patients treated with pancreatectomy without VR and those treated without resection (palliation).
Survival of 22 patients who underwent pancreatectomy with VR was compared with two control groups: 54 patients who underwent pancreatectomy without the need for VR and 28 patients whose pre-operative imaging suggested resectability but whose laparotomy indicated inoperability.
A slight survival benefit was noted in patients who did not require VR (33.5%) compared to those who did require VR [20%, p = 0.18], although not reaching statistical significance. Despite a low 15% three-year survival in patients treated palliatively, this was not statistically different compared to survival after resection with VR (P = 0.23). The presence of nodal metastasis was associated with worse survival (p = 0.006), and the use of adjuvant therapy was associated with better survival (p = 0.001).
Pancreatic cancers that require VR to completely resect the tumor have a similar survival to those not requiring VR. Long-term survival was achievable in approximately 1 out 5 patients requiring VR, although we were not able to demonstrate statistically improved survival compared to palliative care.
KeywordsPancreatic cancerVascular reconstructionPost operative survival
Pancreatic cancer is one of the leading causes of cancer-related deaths in the USA, second only to colorectal cancer as a cause of gastrointestinal-related death.1 More than 37,000 new cases are expected to be diagnosed in 2007 with mortality in excess of 33,300.1 The overall 5-year survival rate has remained <5% for the last three decades despite all the recent advances in diagnosis and staging. Treatment of pancreatic cancer includes multiple modalities but surgical resection is the only potentially curative treatment. Unfortunately, because of the late presentation of the disease, only 15 to 20% of patients are considered for pancreatectomy. Although surgery offers the only potential chance for long-term survival, the prognosis in most patients is poor even in those with resectable disease, where the 5-year survival remains less than 20%.2–4
Approximately 40% of patients with pancreatic cancer present with locally advanced nonmetastatic disease. Tumor adherence or invasion into adjacent structures, particularly the celiac and superior mesenteric vasculature (T4 or stage III disease) makes complete resection very difficult or impossible. Portal or superior mesenteric vein involvement with the tumor has previously been staged as T4,5 and therefore, is considered a contraindication to surgery in most cases. However, for the last several decades, pancreatic surgeons have been able to resect and reconstruct either or both of those veins resulting in a reclassification by the American Joint Commission on Cancer (AJCC) of tumors invading the mesenteric veins at T3.6 In a large single center series published by Tseng et al. at M.D. Anderson Cancer Center,7 a total of 141 patients underwent pancreaticoduodenectomy (PD) with vascular reconstruction (VR). The authors concluded that patients undergoing VR had survival similar to those who underwent standard PD (median of 2 years) and superior to historical patients with locally advanced disease who were treated non-operatively.
The aim of this study is to confirm whether patients with locally advanced pancreatic cancer and mesenteric vascular invasion have similar survival with extended pancreatectomy with VR and to compare their survival to patients treated with pancreatectomy without VR. Further, we compared survival for resected groups to unresectable patients. We hypothesized that a proportion of patients requiring VR are curable by surgery but with a prognosis worse than patients without mesenteric vascular invasion.
This minimal-risk study was approved by Mayo Clinic Institutional Review Board. In a retrospective review, data were reviewed on 104 patients with pancreatic adenocarcinoma, who underwent an exploratory laparotomy with the intent of pancreatic resection for adenocarcinoma between February 1998 and February 2005 at Mayo Clinic in Jacksonville, FL. Vein resection was performed if it permitted complete resection of tumor.
Patients underwent standard or pylorus-preserving pancreaticoduodenectomy (PD), distal pancreatectomy or total pancreatectomy, depending on the primary pathology and location. Twenty-two patients underwent VR of either portal vein, superior mesenteric vein (SMV) or both (VR group) because of gross tumor involvement. It should be noted that the surgeons involved usually attempt to dissect the tumor from the vein initially, and only proceed with VR if it appeared there was direct venous invasion.
To maintain a homogenous group of patients, only VRs involving tangential or segmental resection with subsequent venous patching, interposition grafting, or primary anastomosis were included. The remaining patients were categorized into two control groups to separately determine the survival benefit of pancreatic resection and vascular reconstruction. The first control group “palliative care” (P group) included 28 patients with pancreatic cancer whose pre-operative imaging suggested resectability but whose laparotomy identified metastatic disease or arterial vascular encasement that precluded vascular reconstruction. The patients were typically treated with biliary-enteric bypass, followed by palliative therapy. The second control group included 54 patients with pancreatic cancer who underwent an exploratory laparotomy with pancreatectomy without the need for PV or SMV reconstruction (R group). Data on patients’ demographics, treatment, histopathology, and follow-up were recorded. Patients who underwent chemoradiation were confirmed to have completed the treatment courses recommended.
Tumor extension to the superior mesenteric artery (SMA) or celiac axis, as defined by the absence of a fat plane between the tumor and these arteries by CT, MRI or EUS.
The presence of extra pancreatic metastatic disease identified by presurgical imaging studies.
Severe medical comorbidities including oxygen-dependant obstructive pulmonary disease, unstable coronary artery disease, and other uncontrolled malignancies.
The extent of venous involvement by the primary tumor was not a contraindication for operation as long as there was no CT evidence of tumor extension to the celiac axis or SMA, and there was a suitable SMV below and PV above the site of venous involvement. In this study, the patient who underwent previous attempts of pancreatectomy and those with other indications for pancreatectomy including islet cell or neuroendocrine tumors were excluded.
The technique of venous reconstruction generally depended on the length of the venous involvement by the tumor. No heparin was administered. Involvement of the lateral portal or superior mesenteric vein was managed by proximal and distal control, excision of the involved vein, and primary closure. In cases where hemodynamically significant narrowing of the vein would result from primary closure, a vein patch closure was utilized. Cases with >180° of vein involvement usually required segmental resection and primary end-to-end anastomosis. If the segmental resection resulted in tension, or the resected portion was too long for primary anastomosis, a vein graft including composite graft reconstructed with gonadal and inferior mesenteric veins was utilized. The last option was to utilize a synthetic graft (FEP ringed Goretex vascular graft, 14-mm diameter, catalogue number R14030030). All resections and reconstructions were carried out with at least ×2.5 magnification. Cross-clamp time of the SMV and PV was kept to a minimum to avoid edema of the bowel.
All surgical pathology specimens were evaluated by the department of pathology of the Mayo Clinic Jacksonville. Data was reported using as standard the College of American Pathologist (http://www.cap.org/apps/docs/cancer_protocols/protocols_index.html)/AJCC template for pancreatic cancer,8 which includes a description of the size the specimen, histologic grade, regional lymph nodes, pancreatic, bile duct, and gastric, or duodenal margins, radial margins, venous lymphatic invasion, and perineural invasion. A post-surgical staging was provided by the pathologist (TNM system) and verified on all patients. The TNM classification of the 2002, sixth edition of the AJCC6 was used in all cases. In cases before 2002, the gross and microscopic descriptions of vascular involvement were reviewed in each case and restaged using the current 2002 criteria. The status of vein invasion by gross and microscopic criteria was recorded in all patients in the VR group.
All data analyses were performed using Statistical Package for the Social Sciences (SPSS) version 13.0 software (SPSS, Inc., Chicago, IL). Continuous variables were summarized using medians and ranges, whereas categorical variables were summarized using proportions. Because the data were small and not normally distributed, comparisons of patient characteristics were performed using nonparametric tests. Fisher’s exact test was used to compare categorical variables, whereas Kruskal–Wallis and Wilcoxon rank–sum tests were used for continuous variables in the univariate analysis.
Survival and follow-up were calculated from the time of exploratory laparoscopy to date of death or last available follow-up. All deaths from any cause were included in the survival analysis and subsequent multivariate analysis. Overall survival was demonstrated using the method of Kaplan and Meier. Log-rank test was used to evaluate differences between survival curves. All differences and associations were considered significant at two-sided P < 0.05.
Univariate and multivariate analyses determining the effects of potential prognostic factors on survival were done using log-rank test and Cox proportional hazards analysis. To assess the differences among three groups, two pair-wise comparisons were performed. The first comparison was performed between the patients who underwent pancreatic resection with vascular reconstruction and those who underwent pancreatic resection without VR. Covariates included VR, advanced T stage, nodal metastasis, Eastern Cooperative Oncology Group (ECOG) status,9 pancreatic resection margin involvement, and adjuvant therapy. The second comparison was between the patients who underwent resection with VR and those without pancreatic resection. Covariates included pancreatic resection, nodal metastasis, distant metastasis, ECOG status, and adjuvant therapy.
Summarizes the Demographics, Tumor Staging (TNM status), and Type Surgical Resection in All 76 Patients Who Underwent Pancreatectomy, and 28 Patients Who Underwent Exploratory Laparatomy
R Group (n = 54)
VR Group (n = 22)
P Group (n = 28)
Median age (range)
Advanced T Stage (T 3–4)
Type of surgery
Adjuvant (or palliative) therapy
Type of VR in All Cases
Type of Surgery (N)
Total pancreatectomy (2)
Distal pancreatectomy (1)
Final Surgical Pathology in All Three Groups
Post Surgical Pathology
R Group (n = 54)
VR Group (n = 22)
P Group (n = 28)
Adenocarcinoma (ductal origin)
Acinar cell variant
IPMN with invasive carcinoma
PV or SMV tumor invasion was confirmed in 14 out of 22 cases of VR (64%) by gross or microscopic pathologic examination. The depth of invasion ranged from adventitial to transmural invasion. No invasion was noted in 5 out of 22 cases (23%), and no data could be obtained about vessel invasion in 3 out of 22 cases (14%).
None of the 76 patients who underwent resection died within 30 days of surgery, whereas two patients from the palliative group expired during the same period of time. A total of 11 surgeons were involved in pancreas resections within the study timeframe. Out of those, only five performed the VR, with one surgeon performing 45% of the overall VR. Intraoperative blood flow was assessed clinically at the completion of reconstruction, and color-flow Doppler ultrasound was performed during the post-op period to assess patency of the reconstructed veins. No patient was found to have superior mesenteric or portal venous thrombosis after reconstruction.
The Estimated Survival with Corresponding 95% Confidence Intervals
Estimated Patient Survival (95% Confidence Interval)
Number at risk
Multivariate Analysis Using Cox Regression Analysis Comparing R and VR Groups
Hazard Ratio (95% Confidence Interval)
PV or SMV reconstruction
Advanced T stage
Multivariate Analysis Using Cox Regression Analysis Comparing Patients Who Underwent Pancreatectomy (Groups R and VR Combined) and P Group
Hazard Ratio (95% Confidence Interval)
Univariate Analysis of Survival
Type of Comparison
Log–Rank Analysis (P Value)
R group (comparison 1)
P group (comparison 2)
P group (comparison 3)
Pancreatic adenocarcinoma is one of the most aggressive gastrointestinal malignancies with limited long-term survival even in cases undergoing surgical resection with curative intent. Thus, treating clinicians remain reluctant to refer patients for such a surgery that has significant morbidity and mortality.
The first report of SMV resection and reconstruction came from the University of Minnesota by Moore and his colleagues in 1951.10 Subsequent studies established the long-term patency of autologous vein grafts compared to the synthetic prosthesis where occlusion is not uncommon.11 Although Japanese surgeons used this method to improve survival by widening the margins of pancreatectomy to involve surrounding structures,12 it was not until the 1973 when the concept of “en bloc” pancreatectomy was further defined by Fortner.13 These authors speculated that resecting a wider margin of healthy tissue and the involved vessels would translate into survival benefit.
VR was initially performed in an effort to maximize soft tissue and lymphatic excision. However, from subsequent work done by Yeo et al., it became evident that widening the surgical margins to include more lymph tissue had little effects on survival.14 One large series demonstrated that with proper patient selection and surgeon experience, VR can be performed safely with complication rate similar to standard pancreatic resections.7 The low complication rate reported with vascular reconstruction and the improving operative morbidity and mortality after PD makes it reasonable to consider vascular resection to achieve an R0 resection. Despite that, VR at the time of pancreatic resection remains a controversial approach because of the complexity of the surgical procedure itself and the lack of evidence of survival benefit. Our study suggests that approximately one in three patients with mesenteric vascular invasion can achieve long-term survival with surgery including VR. Although the long-term survival after surgery with VR was numerically better than palliative care group, the small number of patients did not allow demonstration of statistically improved survival.
There remains no consensus on the specific indications for vascular resection of the SMV or its confluence with PV. In half of the cases in our study, the decision to proceed with VR was made before the laparotomy, based on evidence for PV or SMV invasion by endoscopic ultrasound (EUS), computed tomography (CT scan), or magnetic resonance imaging (MRI). In a previous series by Tseng et al.,7 this decision was primarily made at the time of surgery where tumor adherence to these venous structures prevented the surgeon from mobilizing the SMPV confluence from the pancreatic head and uncinate process, as is necessary for standard PD.
The need for VR may be due to adherence to vasculature without actual invasion. In this series, pathology of the excised pancreas confirmed vascular invasion in 64% of cases, compared to 61% of cases in the series by Tseng et al.7 Peritumoral inflammation with adherence to surrounding vasculature may necessitate VR even in the absence of pathologically confirmed invasion. This makes pre-operative staging and decision-making more challenging since all current imaging methods (EUS, CT, MRI) have limited ability to distinguish vascular abutment, adherence, and invasion.15 A recent emerging technique involves the use of intravascular ultrasound intraoperatively to assist the differentiation of true vascular invasion versus adherence only. It further suggests that surgeons should be prepared to perform VR regardless of whether the pre-operative staging indicated vascular invasion. In some cases, involvement of the lateral superior mesenteric or portal vein was evident only after the pancreas was transected. At that point, vein resection must be performed to achieve an R0 resection.
The major limitation of this study is the retrospective nature that could result in variations in pre-operative selection criteria and mixed patient populations. We have made specific efforts to minimize these limitations.
The success of pancreatectomy with VR requires careful selection of patients. Only recently has there been an effort to establish criteria for resectable, borderline resectable, and unresectable pancreatic cancer. In our series, we applied strict preoperative criteria for local tumor resectability in line with NCCN Pancreatic Adenocarcinoma Clinical Practice Guidelines for Criteria Defining Resectability Status.8 These include patients with evidence of arterial involvement, aortic or inferior vena cava invasion or encasement, involvement of the SMV below the transverse mesocolon, involvement of long segments of vein, or occlusion of the SMV or PV. Regional lymph node involvement and direct extension of the primary tumor to adjacent organs were not considered contra-indications to resection. Such criteria are necessary to avoid the inclusion of patients with grossly incomplete resections. Survival duration in this group may be affected more by the failure to remove all gross tumor than by other potential prognostic variables. Varadhachary et al.16 have emphasized that survival of patients who do not have an R0 resection is no different from patients with locally advanced unresectable disease. The only exception in our study was the inclusion of one patient with VR, whereby surgical pathology revealed SMA invasion (in addition to known SMV invasion), which was not detected on pre-operative radiographic assessment.
Our study included only patients with pathologically confirmed malignant tumors of the pancreas that minimized differences among the three patient groups (one study and two control groups). In addition, all resections performed in this study were primary, excluding patients who had previous exploration or unsuccessful attempts of pancreatic resections that might increase the likelihood of peritoneal seeding17 or influence the tumor’s subsequent behavior. The fact that most of the resections were performed by two experienced surgeons helps reduce the technical differences in performing the reconstruction between the different surgeons and likely contributed to the absence of operative mortality in this particular group. In addition, we have adequately controlled for co-morbidities by including the ECOG performance status in the analysis of all the groups.
In the analysis of patients who underwent pancreatectomy, we found no statistically significant survival difference between the VR and the two other control groups, after controlling for potentially confounding variables. Although there may be a small survival difference among patients with locally advanced metastatic disease compared to metastatic disease within the P group, we elected to combine these subgroups due the small number of patients and to facilitate the analysis. Despite the fact that long-term survival after surgery with VR was numerically better than palliative care, the small number of patients did not allow demonstration of statistically improved survival.
In summary, we were able to achieve a 3-year survival of 20% in patients requiring VR, although we were not able to demonstrate statistically improved survival compared to the palliative care group. Future studies may be needed to further identify a subgroup of patients who may benefit from this surgical approach. Until then, we suggest that this type of procedure be performed by skilled pancreatic surgeons in high-volume centers.