Reports of the use of vascular resection at the time of pancreatectomy have circulated since Fortner attempted to popularize the techniques approximately 40 years ago. At that time, resection of major mesenteric vessels was advocated as part of a regional pancreatectomy procedure designed to maximize soft-tissue clearance in the face of infiltrating cancers of the pancreas.^1,2 Although vascular resection subsequently lost favor, as it was perceived to be associated with significant morbidity and mortality, the past 20 years have seen a greater understanding of both its role and its limitations. Experienced surgeons have steadily been increasing the use of vascular exposures and resection techniques to achieve margin negative resections as part of comprehensive multimodality treatment strategies for patients with pancreatic ductal adenocarcinoma (PDAC).

In this chapter, the role of vascular resection at pancreatectomy is reviewed within the context of a discussion of “borderline resectable” disease. Controversies with regard to staging and management of borderline resectable PDAC are discussed. Finally, surgical techniques are reviewed and illustrated in detail.

A complete microscopic (R0) pancreatectomy operation has long been considered necessary for long-term survival of patients with PDAC.³ In contrast, incomplete (R2) resection has long been considered a palliative procedure that is not associated with any extension of survival relative to other less invasive strategies.

The likelihood of R0 resection in a patient with PDAC can be estimated prospectively on the basis of cross-sectional computed tomography (CT) studies. Tumors likely to be removed with microscopically negative margins at surgery have historically been considered “potentially resectable” and can be distinguished from patients with “unresectable” tumors—for whom surgery is expected to be associated with positive margins and early treatment failure—on the basis of good imaging.

Over the past several years, however, it has increasingly been recognized that a subset of patients with scans suggesting that R0 resection is unlikely may actually undergo resection and have postoperative survival durations similar to those of patients with resectable cancers. However, such tumors are at high risk for positive microscopic (R1) margins when surgery is attempted de novo, even when surgery is performed by the most skilled of pancreatic surgeons. Patients with these advanced cancers may therefore benefit from preoperative therapy delivered by multidisciplinary treatment teams prior to radical surgical operations that may involve resection of major mesenteric vasculature.^4,5 These advanced tumors, for which potentially curative therapy is a rational consideration, are considered “borderline resectable.”

Pretreatment Evaluation and Management

ANATOMIC (RADIOGRAPHIC) DISEASE STAGING

Anatomic staging of PDAC is routinely accomplished using multidetector CT scanners and a specialized protocol optimized to image pancreatic tumors. Intravenous contrast and a low-density oral contrast are administered, and images are obtained in arterial, portal-venous, and noncontrast phases. Thin slices obtained every 2 to 3 mm are obtained through the pancreas to facilitate the imaging of small tumors and the characterization of the anatomic relationships between the cancer and adjacent mesenteric vasculature. Coronal and sagittal reconstructions may be obtained to clarify the surgical anatomy and to assist in surgical planning.

Endoscopic ultrasonography (EUS) may also be performed for patients suspected to have borderline resectable PDAC. EUS may be used to further elucidate locoregional anatomy but is not superior to CT in this regard, and echoendoscopic images are less appropriate to the surgeon for surgical planning because it does not clearly resolve the anatomy of small venous and arterial vessels that are critically important. However, EUS is useful because fine-needle aspiration (FNA) biopsies may be performed under echoendoscopic guidance.

Staging laparoscopy enhances the detection of occult metastases in patients with radiographically localized disease; extrapancreatic cancer is found in approximately one-third of patients with resectable and locally advanced PDAC staged with cross-sectional imaging alone.⁶ Staging laparoscopy is therefore a rational study for patients with borderline resectable PDAC prior to resection. Staging laparoscopy may also be considered even earlier—prior to the initiation of neoadjuvant therapy—to exclude radiographically occult metastatic disease that may preclude therapy in the potentially curative mode. However, we have historically not staged patients with borderline resectable PDAC routinely prior to the administration of systemic chemotherapy because the identification of metastatic disease would not typically change the plan for systemic therapy. Instead, we have reserved it for the preoperative setting either as a separate procedure or immediately before laparotomy under the same anesthetic.

NONANATOMIC STAGING

Although precise definition of tumor anatomy with cross-sectional imaging is clearly an important aspect of disease staging, assessments of each tumor’s predicted natural history—its “tumor biology”—and each patient’s personal physiology are also critical to personalize treatment strategies and optimize outcomes. Because we consider the presence of adverse biologic and physiologic features to represent rationale against the use of an initial surgical approach even when tumor anatomy appears otherwise resectable, we consider clinical indicators of both tumor biology and patient physiology as additional factors used to stage disease in the MD Anderson Cancer Center classification system of patients with localized PDAC.^7,8 Specifically, borderline resectable type A patients have borderline resectable anatomy (using radiographic definitions described below). Borderline resectable type B patients have either potentially or borderline resectable anatomy along with clinical findings suspicious but not diagnostic for extrapancreatic disease: for example, indeterminate liver lesions, a serum carbohydrate antigen (CA) 19-9 level ≥ 1000 U/mL (with a normal total bilirubin), or biopsy-proven involvement of regional lymph nodes. Borderline resectable type C patients have potentially or borderline resectable anatomy along with advanced age (≥80 years), severe comorbidities requiring extensive evaluation or optimization, or depressed PS (ECOG ≥ 2). Although these defined clinical classifications have not been widely adopted, they are an invaluable way to both convey prognosis among healthcare providers and to optimize therapeutic plans for individual patients.

BILIARY DRAINAGE

Routine biliary drainage is not recommended for most patients scheduled to undergo surgery for PDAC because drainage may increase the rate of serious adverse events.⁹ However, biliary drainage is required for patients who present with or develop biliary obstruction from borderline resectable disease in whom neoadjuvant therapy is either planned or in progress, because many of the cytotoxic regimens and agents used in the preoperative period are cleared by the biliary system.¹⁰ Metal stents are preferred over plastic stents on the basis of improved patency rates (median patency times 3.6 vs. 1.8 months).¹¹ The shortest stent required to bridge a biliary obstruction should be used. No data exists to support or refute the role of prophylactic stent placement prior to the initiation of therapy in a patient with borderline PDAC who does not present with jaundice (i.e., from a tumor of the uncinate process), but stent placement should be considered early in all of these patients with infiltrating disease.

TISSUE BIOPSY

Acquisition of a tissue biopsy is mandatory prior to the initiation of preoperative therapy for borderline resectable PDAC. Cytological samples can be obtained by brushing a biliary stricture at endoscopic retrograde cholangiopancreatography (ERCP), or cells can be obtained with FNA either at EUS or under CT or US guidance. EUS is favored over percutaneous imaging because the biopsy needle tract passes through the duodenum, which is removed at PD; this reduces the remote but possible risk for recurrence/progression due to seeding of the needle tract with cancer cells.¹² Small-diameter biopsy instruments may be used to acquire larger core tumor samples to examine tissue architecture or for research purposes.^13,14 Concerns over inducing pancreatitis or bleeding have limited the routine use of core biopsy techniques to date but their use is becoming more common.

Radiographic Definitions of Borderline Resectable PDAC

Although relative consensus exists that the term “borderline resectable PDAC” should be used only to describe patients at high risk for margin-positive resection and subsequent treatment failure when surgery is performed de novo, no consensus yet exists on what set of radiographic features is unique to those tumors.

The first collection of radiographic criteria used to define a subset of localized PDAC tumors with anatomic features intermediate between resectable and unresectable appears to have been first described in a retrospective study published in 2001.¹⁵ In that report, 15 patients who had “marginally resectable” PDAC—defined as tumors in which the perivascular fat plane was absent over 180 degrees of the superior mesenteric artery (SMA), superior mesenteric vein (SMV), or portal vein (PV), and persisted for a length of greater than 1 cm—were treated with 50.4 to 56 Gy external beam radiotherapy (EBRT) with concurrent 5-fluorouracil (5-FU) in an attempt to improve resectability and local control. As the potential benefits of neoadjuvant therapy with regard to both patient selection and margin sterilization were becoming more understood, and as the use of venous resection at pancreatectomy became more widespread, interest in the use of similar potentially curative strategies for locally advanced cancers increased. By 2006, the National Comprehensive Cancer Network (NCCN) used the term “borderline resectable” to describe localized cancers thought to be at high risk for treatment failure when treated primarily with surgery and suggested new radiographic criteria to identify them. Investigators at MD Anderson Cancer Center,¹⁶ Fox Chase Cancer Center,¹⁷ and elsewhere subsequently built on this evolving framework.

Through this 15-year history, several obstacles have stood in the way of widespread adoption of a uniform set of radiographic criteria to define borderline resectable disease. Indeed, at least three primary radiographic definitions for “borderline resectable” PDAC currently exist in the literature, but most studies that report on patients treated for “borderline resectable” PDAC do not use any one of them and instead adopt some proprietary or modified set of radiographic criteria¹⁸ (Table 142-1).

First, consensus has not yet been achieved on a single histopathologic protocol to examine and report the status of surgical margins following pancreatectomy. Indeed, differences in the methods in which surgical specimens are handled, inked, evaluated, and reported contributes to significant variability in terms of margin-negative rates: although R1 rates from 20% to 40% are routinely reported in single-institution series and clinical trials, cells can be identified in the margins of as many as 90% of resected specimens when a rigorous protocol is used.^19–21 Absence of consensus on this surgical endpoint makes series from different centers difficult to compare, and thus the results of therapies and operations almost impossible to interpret.²²

Second, strong biases continue to exist with regard to the use of vascular resection at PD, the administration of neoadjuvant therapy prior to surgery, and the role of potentially curative operations for patients with PDAC in general. To the extent that the distinction between resectable and borderline resectable is often used to discriminate between tumors that should be treated first with nonoperative therapies and those which should be treated with resection, the label of borderline resectable is often used to rationalize the use (or disuse) of nonoperative therapies in the context of these biases.¹⁶ Indeed, decisions on whether preoperative therapy should be administered or not are often individualized on the basis of perceived technical difficulty of the case or other nebulous clinical factors—only patients with disease felt to warrant preoperative therapy in the context of these biases are labeled “borderline resectable.” The radiographic criteria used to define a group of tumors labeled “borderline resectable” therefore often vary between or even within institutions.

Finally, the confusing descriptive lexicon utilized in radiographic reports makes description of radiographic findings difficult to interpret and compare. Multiple subjective qualifiers have been used to describe radiographic findings perceived to indicate vascular involvement. Commonly used descriptions of the tumor–vessel relationship such as abutment, impingement, and encasement are ambiguous. Descriptions of the effects of the tumor on the vessel, such as narrowing, shift, and distortion, are not quantitative and represent a broad spectrum of actual vessel involvement. The radiographic anatomy of tumor and vessel may be more simply and objectively described in terms of the degree of circumferential interface between tumor and each adjacent mesenteric vessel.^23,24

In each of the major radiographic definitions, radiographic evidence for minimal involvement of arterial structures distinguishes borderline resectable from locally advanced disease because (1) the celiac artery and superior mesenteric artery (SMA) cannot routinely be resected due to high morbidity and mortality,²⁵ (2) a radiographic interface between tumor and artery measuring >180 degrees suggests the presence of tumor that cannot be resected without resection of the artery,²³ and (3) neoadjuvant therapy has not been successful in “downstaging” locally advanced tumors to more easily resectable ones.^26,27 However, more radiographically advanced hepatic arterial involvement suggesting need for hepatic artery resection is considered borderline resectable because resection and reconstruction of that artery may be appropriate in certain clinical situations.²⁸

The extent of the interface between tumor and vein that distinguishes resectable from borderline resectable disease is more controversial. The MD Anderson Cancer Center definition considers any tumor that does not lead to occlusion of the vein as resectable; tumors with reconstructable occlusion are considered borderline resectable and tumors with unreconstructable occlusion are considered locally advanced. Tumors with a smaller extent of vein involvement are considered borderline resectable on the basis of the National Comprehensive Cancer Network (NCCN) and Society for Surgery of the Alimentary Tract (SSAT)/Americas Hepato-Pancreato-Biliary Association (AHPBA)/Society of Surgical Oncology (SSO) definitions.²⁹ All of these existing definitions have attempted to classify tumors on the basis of their radiographic anatomy and have therefore been necessary to drive the field forward to some degree, because they have distinguished these cancers prospectively, not retrospectively—as some continue to do.^30,31 However, these definitions themselves are not entirely evidence based and so they have also contributed to significant confusion.

Evidence does exist to help prospectively and radiographically define a group of pancreatic cancers at high risk for R1 resection if surgery is utilized de novo. In an early study of 35 patients who underwent venous resection at pancreatectomy without prior therapy, patients with bilateral narrowing or long-segment abutment of the SMV–PV on preoperative angiograms had a poorer prognosis than patients with radiographically normal, shifted, or unilaterally (right) narrowed veins.³² More recently, in a study of 463 patients who underwent pancreatectomy among whom 297 patients required venous resection, long-term survival was identified only in patients with unilateral narrowing or normal veins.³³ In another examination of 375 patients who underwent resection of pancreatic cancer, R0 rates of 77% and 76% among patients with normal or unilaterally narrowed veins, and 64% and 65% among patients with veins that were bilaterally narrowed or occluded were observed. Tumors that had an interface with the common hepatic artery (CHA) or SMA were associated with R0 rates of 48% and 37%, respectively.³⁴ Together, these data suggest that R1 resection and poor survival are more common when tumors infiltrate to the left lateral aspect of the vein.

Based on these data, investigators enrolling patients to a multi-institutional trial of neoadjuvant therapy for borderline resectable disease have proposed a definition of borderline resectable pancreatic cancer.¹⁸ The definition uses objective terminology and is applicable using routine, axial pancreatic protocol images. The definition includes tumors radiographically as localized cancers with one or more of the following: (1) an interface between the primary tumor and SMV–PV measuring 180 degrees or greater of the circumference of the vein wall, and/or (2) short-segment occlusion of the SMV–PV with normal vein above and below the level of obstruction that is amenable to resection and venous reconstruction, and/or (3) short-segment interface (of any degree) between tumor and hepatic artery with normal artery proximal and distal to the interface that is amenable to resection and arterial reconstruction, and/or (4) an interface between the tumor and SMA or celiac trunk measuring less than 180 degrees of the circumference of the artery wall. Although the NCCN has not endorsed this definition completely, it has been endorsed by the NCCN for all trials going forward for borderline resectable cancer.³⁵

Neoadjuvant Therapy

RATIONALE

Because borderline resectable tumors are at high risk for incomplete (R1) resection following surgery de novo, consensus exists that they should be treated initially with both systemic chemotherapy and chemoradiation.³⁶ The administration of preoperative therapy to patients with these tumors is associated with important potential advantages. The administration of nonoperative treatments prior to surgery ensures that all patients who undergo resection receive them. This is critical because adjuvant therapy—which has been clearly established to improve survival relative to surgery alone—is received in only about half of patients who undergo resection for PDAC in general, and its receipt is inhibited by both perioperative complications and tumor progression.^37,38 The early delivery of systemic therapy is perceived to be particularly important as most patients present with ostensibly localized PDAC have micrometastatic disease.³⁹ Chemoradiation is also thought to be beneficial because the celiac axis and SMA are encased in a network of nerves that is believed to facilitate tumor extension.⁴⁰ Chemoradiation may sterilize the surgical margins adjacent to these vessels and lead to more durable local and distant control.^41,42 The time during which these therapies are administered may also be beneficial: the interval before surgery allows for rapidly progressive disease to declare itself. Patients who have progressive disease through therapy are presumed to have “aggressive tumor biology” that would not have responded well to surgery at any point following presentation.