Limits of surgery for pancreatic cancer

Summary

Pancreatic cancer patients often present in an already advanced state of disease and the therapeutic approach is an interdisciplinary challenge. Surgery is an integral part in a potentially curative setting, yet in such advanced disease surgery can reach its limits. The technical feasibility has to be weighted against potential harms and the oncological reasonableness.

In locally advanced disease, limits of surgery could be pushed as evidence grew. A venous vascular tumor infiltration nowadays does not preclude patients from surgery, as venous resections can be safely performed and survival rates are not inferior to patients with standard resections. Multivisceral resections have an increased risk of morbidity and mortality, but can improve overall survival. The resection and reconstruction of tumor infiltrated arteries is technically feasible, but these procedures have a high rate of associated morbidity and mortality with an unclear oncological benefit and therefore are generally not recommended. This also holds true for intentional palliative R2-resections, which do not offer a survival benefit but decrease the quality of life and have higher morbidity and mortality rates than palliative bypass procedures. A synchronous resection of the primary tumor and intraabdominal metastases in an olgiometastatic disease only offers a questionable oncological benefit and the evidence for this approach is scarce. Therefore, surgery in a metastatic disease is generally not recommended and has to be discussed interdisciplinary on a highly individual basis.

Key words:
pancreatic cancer − multivisceral resection − staging

Introduction

For decades, the pathogenesis of pancreatic cancer (PDAC) as well as diagnostic and therapeutic approaches have been the focus of basic science and clinical research. Though knowledge has accumulated over the years, the prognosis for PDAC generally remains poor. With more than 100 000 deaths in the USA and Europe per year, PDAC is the fourth leading cause of cancer related deaths in the Western world [1,2]. Efforts have been made to improve survival by adding adjuvant systemic therapies and, more recently, by investigating multi-modality treatments before a surgical resection. Thereby, the cornerstone of any therapy is still a radical surgical resection, yet only about 20% of all patients present with resectable disease. Following surgical resection, the median survival rates range between 11–20 months with 5-year-survival rates of 7–25%, though local recurrence as well as (mostly) hepatic metastasis formation within 12 to 24 months account for treatment failures. Median survival in locally advanced or metastatic disease is even more reduced, with 10–12 months and 4–6 months, respectively [3,4].

As mentioned, the only potentially curative approach remains in surgical resection. Therefore, one of the achievements in surgical therapy was to push the limitations of resectability – but technical feasibility and concomitant morbidity and mortality have to be carefully weighted against each other. This also holds true for unintentional or intentional R1/R2-resections or resection of synchronous liver metastases. These issues will be addressed subsequently.

Definitions of borderline resectable disease

One of the prerequisites of a surgical resection is the assessment of resectability, mainly derived from abdominal CT-scans. In the past, the definitions of resectable and borderline resectable disease were nonuniform. Recently, an updated German national guideline on pancreatic cancer has been published which adopted the American NCCN definitions on borderline respectability [5]. These expert consensus statements define tumors as localized and resectable in case of

• no distant metastases
• no radiographic abutment, distortion tumor thrombus or venous encasement of the superior mesenteric vein (SMV) and the portal vein
• clear fat planes around the superior mesenteric artery (SMA), the coeliac axis and the hepatic artery [6].

In these cases, a surgical resection seems possible without the need of vascular or extended resections, though, of course, the situation can present differently during surgery (e.g., peritoneal carcinomatosis, small liver metastases, vascular tumor infiltration). Yet, in general, these tumors don’t confront surgeons with the limits of pancreatic surgery.

The NCCN definitions were also confirmed in a consensus statement by the International Study Group on Pancreatic Sugery (ISGPS) and are cited as follows:

• „CT findings of venous distortion of the SMV/ portal venous axis even including short-segment venous occlusion with proximal and distal sufficient vessel length allowing safe reconstruction.
• Encasement of the gastroduodenal artery (GDA) up to the hepatic artery, with either short-segment encasement or direct abutment of the hepatic artery without extension to the celiac axis.
• Tumor abutment of the SMA but with no greater than 180° of the vessel wall circumference.

Whereas 180° encasement of the GDA is still considered borderline resectable, the encasement of the SMA must be no greater than 180° for borderline resectability, regardless of whether the tumor is in the head, body, or tail of the pancreas. Regarding the celiac axis and tumor location, a tumor is considered to be unresectable if there is any celiac abutment from a tumor in the pancreatic head and also unresectable if there is celiac encasement of greater than 180° from a tumor in the body or tail.” [7]

This subgroup of patients is in special focus, as surgical strategies as well as neo-adjuvant therapy regimes aim to increase the (R0) – resectability.

En-bloc vascular resections in locally advanced pancreatic cancer

As described by the definitions of borderline resectable disease, locally advanced pancreatic tumors tend to infiltrate or encase vascular structures, especially when the tumor is located in the pancreatic head. In the past, a vascular tumor infiltration contraindicated a surgical resection. During time, this dogma has changed. Several studies have compared the outcome of patients who underwent vascular resections to patients in whom no vascular resection was necessary. Evidence accumulated in the 1990’s, that venous vascular infiltration does not preclude patients from surgery and that survival is not inferior to patients without venous tumor involvement. In a report by Fuhrman et al., no differences in either the median hospital stay, morbidity or mortality was observed between 36 patients without venous resection and 23 patients with en bloc resection of the porto-mesenteric confluence [8]. A series of pancreaticoduodenectomies by Leach et al., including 44 patients without vascular resection and 31 patients with resection of the venous confluence, showed equal median survival rates between the two groups (22 months for vascular resected patients and 20 months for the control group)[9]. Similar results were reported by Harrison et al., who compared the survival of 58 patients with segmental portal vein resection or venorraphy to the control group. The median survival was 13 months vs. 17 months, without significant differences [10]. This equality in survival was also confirmed in several reports in the last decade [11−16]. A series by Tseng et al. included 110 patients with vascular resections and also found no difference in median survival compared to the control group (23.4 vs. 26.5 months), even though this series also included a small number of arterial resections. Of interest, the higher rate of R1 resections in the vascular resection group (22%) compared to the lower R1 rate in the control group (12%) did not translate into a survival difference [17]. In the analysis of our own data, the median survival of 100 patients who underwent vascular resection was 15 months compared to 16 months in the control group. Remarkably, in 23 of the 77 patients who underwent vascular resection, histo-pathologic examination did not confirm vascular invasion – attributable to a tumor infiltration mimicking peritumoral inflammation [18]. In a literature review by Siriwardana et al., including 52 manuscripts and 6.333 patients, out of which 26% underwent synchronous venous resections, the morbidity and mortality rates in vascular resected patients were 42 and 5.9%, respectively, the median survival was 13 months [19].

To date, the resection of the superior mesenteric or the portal vein is considered as a standard approach to achieve a complete tumor resection.

In contrast to venous tumor infiltration, tumor infiltration of the celiac axis or the superior mesenteric artery (SMA) is regarded as a sign of aggressive tumor biology and the indication for a tumor resection with combined arterial resection is highly individual [1]. Several reports have been published on the results of pancreatic resections including arterial resections, all presenting rather small patient numbers. A literature review by Mollberg et al. in 2011 included 366 patients who underwent different kinds of pancreatic resections, e. g., duodenopancreatectomy, distal pancreatectomy or total pancreatectomy with arterial vascular resections and 2243 patients without resection. The group of patients who underwent an arterial vascular resection had a significantly increased perioperative risk of mortality (Odds ratio= 5.04, 95% confidence interval 2.69−9.45) as well as a poor survival at 1 and 3 years. These inferiorities for the group of arterial resected patients remained when compared to patients who underwent venous resections [20]. In our own data, the median survival between patients with and without arterial vascular resection was non-significantly different with 14.0 vs. 15.8 months and better than for patients who only underwent palliative bypass procedures. Yet, the increased rates of post-operative morbidity and mortality in arterial vascular resected patients had a striking impact on the suspected oncologic benefit [21].

Though an en-bloc arterial vascular tumor resection per se is technically feasible, these resections are generally no standard choice of treatment due to the high morbidity and mortality rates together with the unsure oncological benefit [5]. This holds true for resection of the SMA as well as the celiac axis or the hepatic artery. In left-sided pancreatectomy, resection of the celiac trunk is reported to possibly be a beneficial option, but has not gained clear evidence yet [1,22]. Figures 1a and 1b show the reconstruction of the left-sided hepatic artery by a saphenous vein graft originating from the right common iliac artery.

In the upcoming ISGPS consensus statement on borderline resectable pancreatic cancer, a straightforward arterial resection as well as a neo-adjuvant therapeutic approach in these cases is only recommended in clinical trials. In patients who were staged as borderline resectable because of features of arterial involvement seen on CT-scan, an abdominal exploration is recommended to verify this suspicion [7].

Multivisceral resections

As achieving a R0-resection is a main goal in PDAC-resection, local tumor extension necessitates a multivisceral resection in about every third patient [1]. These multivisceral resections can include an en-bloc resection of the adrenal gland, the stomach, the mesocolon or colon, the liver and seldom the small bowel. Several publications evaluated the oncological benefit and the technical feasibility of multivisceral resections. In their analysis of 101 patients who underwent multivisceral resections, Hartwig et al. found a median survival rate of 19.8 months for multiviscerally resected patients which was comparable to the control group of patients without these extended resections [23]. Though the survival rates were comparable, the multiviscerally resected group of patients had a higher rate of morbidity and a longer Intensive Care Unit (ICU) stay. Thirty-day and in-hospital mortality were higher, but without reaching statistical significance. The analysis of our own data showed a median survival in multiviscerally resected patients of 16 months, being slightly inferior to standard resections with 18 months but far higher than for palliative bypass procedures. Yet, post-surgical morbidity was also increased in our series [24]. A recently published article that used data from the prospective National Surgical Quality Improvement Project (NSQIP) in the USA included 9,927 cases; 273 of these (3%) were multivisceral resections. In their analysis of this large dataset, the overall morbidity was higher in the multiviscerally resected patient group with 65.2% compared to 39.8% for patients who underwent a standard resection. Higher morbidity rates included cardiac, pulmonary, throboembolic and wound complications as well as renal failure, sepsis, shock and bleeding complications. One independent predictor of morbidity was a performed colon resection. The matched mortality rate was 7.7% for multiviscerally resected patients compared to 3.5% for the control group. Though the dataset used lacked long-term survival, this analysis confirms an increased risk of post-surgical morbidity and mortality [25].

As stated in the national German guidelines for PDAC, the necessity of multivisceral resections is no contraindication for surgery [5]. The survival rates are far higher following a multivisceral resection than following palliative bypass procedures, but the higher morbidity and mortality rates have to be considered and discussed with the patient.

Intentional palliative and unintentional R2-resections

Because of the poor patient survival in advanced pancreatic cancer, alternatives to only palliative systemic therapy have been sought. Though the main goal of a surgical resection is a complete tumor removal and tumor debulking is only integrated in the treatment of selected entities such as ovarian or renal cancer [26,27], some arguments would theoretically favour also a tumor debulking in PDAC. The impact on survival of systemic therapies is not striking and patients with advanced disease often suffer from symptoms related to the tumor (such as pain, biliary and duodenal obstruction) which might be reduced by tumor debulking. Additionally, sometimes an intended curative resection ends up as a R2-resection [28]. In a recent systematic review by Gillen et al., the authors included 4 studies and a total of 138 patients who were compared to a control group of patients who underwent palliative bypass surgery. In the pooled survival analysis, the median survival for patients who underwent a R2-resection was 8.2 months and 6.7 months for double loop bypass patients. The post-surgical morbidity and mortality risk was significantly higher in the pooled group of patients who underwent R2-resections. Patients who underwent R2-resections had an estimated longer hospital stay of 5 days compared to the double bypass group.²⁸ Recovery of quality of life as well as the clinical condition at discharge was better in patients who underwent a double bypass [29,30]. Our own data, which were included in the review, demonstrated an increase in morbidity but comparable mortality for unintentional R2 resections with a significant survival benefit [31]. Our data on intentional R2-resections showed higher morbidity and mortality rates without significant survival benefits [32].

As morbidity and mortality are increased, the quality of life is reduced and the oncologic benefit is small and arguable, an intended palliative tumor resection is not recommended.

Synchronous resection of metastatic disease

Metastatic disease in pancreatic cancer at presentation is common and mainly occurs in the liver, the lung or the peritoneum. In a metastasized setting, surgical options are limited. If the tumor has spread extra-abdominally, there is no benefit to be expected from a surgical tumor resection. This is also true for patients with hepatic and/or peritoneal metastasis at presentation - these patients should receive systemic therapy. The group of patients who might benefit from a resection of the primary tumor and the metastases would be those with an oligo-metastasized disease, e. g. limited and “easily” resectable liver metastases and/or resectable local peritoneal spread discovered during abdominal exploration. Only few reports with small patient numbers exist on these synchronous metastasis resections. In one series by Gleissner et al., 22 patients with periampullary cancer or PDAC who underwent simultaneous resection of liver metastases were analyzed. The median survival of these patients was 5.9 months and 5.6 months in the control group of bypass patients [33]. In another report by Seelig et al. on 14 patients with synchronous liver metastasis resection, the median survival was 11 months [34]. Shirikhande et al. retrospectively assessed a total of 29 patients, who underwent synchronous primary tumor and metastasis resection; 11 patients out of this group had a concomitant liver metastasectomy and median survival was 11.4 months for the R0/R1M1 group compared with 5.9 months for the M1 group who only underwent surgical exploration.³⁵ Addititonally, better outcomes have been reported for synchronously resected ampullary carcinomas than for PDACs [36]. A review from 2008 by Michalski et al. extracted data from 3 publications on synchronous liver metastasectomies, each with more than 10 included patients. The median survival was reported to be between 5.8 and 11.4 months and the authors concluded that overall survival in cases with one or few liver metastases seemed to be comparable to patients without metastatic disease and therefore, the value of liver metastasectomy should be evaluated in clinical trials [37].

Generally, in metastatic disease, the synchronous resection of liver metastasis and the primary tumor is not recommended and has to be interdisciplinary discussed in highly selected patients. In our own routine, patients who present with a solitary liver metastasis will receive neo-adjuvant therapy and will be re-assessed after two to three months of therapy. Thereby, the biological behaviour can be evaluated - if the metastatic situation remained stable and the primary tumor is resectable, we would opt for a synchronous resection of both the primary and the solitary liver metastasis. In case of a previously unknown solitary liver metastasis discovered during surgery, we would also favour a synchronous resection of both the primary and the metastasis, though the evidence for these synchronous resections is vague.

prof. Dr. J.R. Izbicki, FACS, FRCS Ed. Hon.

University Medical Center Hamburg-Eppendorf

Department of General,

Visceral and Thoracic Surgery

Martinistrasse 52

20246 Hamburg, Germany

e-mail: izbicki@uke.de