Imaging

Imaging is now the mainstay in diagnosis and staging of pancreatic tumors, unlike the traditional approach of surgical exploration and a hands-on intraoperative examination to determine resectability.

Helical computed tomography (CT) scanning has been established as the single most effective initial staging study (Freeny, 2001; Klauss et al, 2008) and is often used as the entry point to a management algorithm. The findings then dictate the subsequent management, including additional diagnostics (see Chapter 16).

The previous selling point of magnetic resonance imaging (MRI) (see Chapter 17) was its being a one-stop modality that combines pancreatography, cholangiography, and angiography and also assists in the evaluation of tumors. However, MRI is being progressively overshadowed with the advent of the multidetector CT, which has nearly the same offerings. As such, MRI offers no added benefit over CT for evaluating pancreatic cancer, except perhaps in patients in whom a contrast-enhanced CT cannot be performed (White et al, 2003), and MRI seems to be superior for the differential diagnosis of pancreatic cystic lesions (see Chapter 57).

MR cholangiopancreatography (MRCP) offers a noninvasive delineation of the pancreatic and biliary ducts. It detects pancreatic or ampullary carcinoma by demonstrating the effect of a space-occupying lesion on the ducts, namely obstruction or displacement (Fig. 62A.2). The classic feature is the “double-duct” sign. However, even a strictly defined double-duct sign is only 80% to 85% specific for malignancy (Menges et al, 2000). More recent applications include secretin-enhanced MRCP, which can improve pancreatic duct and side-branch delineation. Furthermore, such pharmacologic stimulation of pancreatic juice secretion can potentially allow the evaluation of pancreatic flow dynamics and assessment of pancreatic exocrine function (Kalra et al, 2003).

FIGURE 62A.2 A, Magnetic resonance image showing a dilated pancreatic duct (arrow) secondary to a small tumor. B, Magnetic resonance cholangiopancreatography shows a clearer delineation of the pancreatic duct. It demonstrates a normal pancreatic duct in the pancreatic head (arrowheads), a filling defect in the pancreatic body, and distal dilation of the pancreatic duct in the tail with appearance of first- and second-order side branches (arrows). C, Specimen photograph. The cause of the ductal obstruction was later shown to be a T1 pancreatic ductal adenocarcinoma (arrow) in the final histology.

Endoscopic ultrasonograpy (EUS) allows a sonographic transducer to be placed in close proximity to the pancreas (see Chapter 14). In doing so, interference from overlying bowel gas is eliminated, thus allowing higher frequencies to be used and resulting in markedly improved resolution of the imagery. Consequently, EUS is more sensitive in detecting smaller lesions (<20 mm), with a sensitivity in the range of 93% to 100% (Tamm et al, 2003). In a meta-analysis of studies comparing staging by EUS to results with other modalities, EUS without fine needle aspiration (FNA) more accurately predicted T stage, N stage, and PV involvement than did CT scanning. EUS also allows image-guided FNA of the primary tumor and the regional lymph nodes without the possible risk of tumor seeding along the needle tract, unlike FNA via the percutaneous route (Gress et al, 2001).

EUS-guided FNA biopsy has been reported to have a high sensitivity (93%) and specificity (100%) (Wiersema, 2002); however, EUS with FNA is only of diagnostic value if histology confirms a pancreatic tumor. The major limitations of this technology are operator dependence and a limited field of visualization for the detection of distant metastasis. In fact, in resectable pancreatic cancer, but also in cystic or endocrine tumors, preoperative biopsies are of no value (Hartwig et al, 2009a). Therefore, we use this modality on a highly selective basis, such as in obtaining tissue biopsy for patients scheduled for neoadjuvant therapy with locally advanced pancreatic cancer, or before palliative chemotherapy is applied in metastatic disease.

With the advent of MRCP, EUS, and multidedector CT with multiplanar three-dimensional (3D) reconstruction, the role of endoscopic retrograde cholangiopancreatography (ERCP) as a diagnostic tool has become increasingly limited. Furthermore, it has potentially high complication rates compared with other procedures (Ciocirlan & Ponchon, 2004). Such concerns have been summarized in a recent report by the American Society of Gastrointestinal Endoscopy (ASGE, 2003). In light of these data, the current view is that purely diagnostic uses of ERCP should be limited, a view affirmed by the National Institutes of Health (NIH) State of the Science Conference Statement (Cohen et al, 2002), which recommended that for patients with pancreatic or biliary cancer, the principal advantage of ERCP is palliation of biliary obstruction when surgery is not elected.

Perioperative Anticoagulation

Meta-analysis of the role of low-molecular-weight heparin (LMWH) in the prevention of venous thromboembolic events in general surgery has shown that LWMH can significantly reduce the incidence of asymptomatic deep vein thrombosis (DVT), clinical venous thromboembolism, and pulmonary embolism, with a trend toward a reduction in overall mortality rate (Mismetti et al, 2001). Consequently, we routinely administer LMWH in a prophylactic dose to our patients, starting the evening before the day of surgery, until the patient is ambulant postoperatively. In addition, our patients are prescribed compression stockings to wear intraoperatively and for their entire inpatient stay because such stockings reduce pooling of blood in deep veins by mechanically preventing venous distension (Morris & Woodcock, 2004) and are thus a simple and cost-effective method of DVT prophylaxis.

Antibacterial Prophylaxis

Antibacterial prophylaxis is instrumental in the reduction of infection-related morbidity with clean-contaminated procedures (de Lalla, 2002); as such, it is recommended for all patients who undergo hepatobiliary or pancreatic surgery (Sganga, 2002). Drugs with antianaerobic activity are added if anaerobe encounter is anticipated during the procedure, in particular procedures involving the gastrointestinal (GI) tract. The general guideline is to use the highest licensed dosage of the chosen antimicrobial agent, administered at induction of anesthesia so as to achieve high peak tissue concentration at the site of the wound before the first incision, which should be maintained until the time of closure (Polk et al, 2000). For our pancreatic operations, in addition to 500 mg metronidazole, we routinely give 4 g mezlocillin, an ureidopenicillin with activity against both gram-positive and gram-negative organisms, including enteric bacilli. We generally do not give postsurgical doses unless the patient has had a recent bout of cholangitis or if the bile was found to be turbid or purulent intraoperatively. In any event, in all procedures that require entry into the biliary tract, bile is sent for microbiologic examination to guide postoperative antimicrobial treatment if the need arises.

Octreotide Analogues

The pancreatoenteric anastomosis is considered the Achilles heel of PD, a testimonial to the potentially disastrous sequelae of life-threatening intraabdominal sepsis and hemorrhage (Stojadinovic et al, 2003) in the event of a pancreatic leak. The secretion capacity of the remnant gland has been found to be a determining factor for the development of a leak because continuous pancreatic secretion has been hypothesized to hinder healing of the pancreatic stump. This led to the hypothesis that by reducing exocrine secretion, perhaps the incidence of pancreatic fistula could be reduced accordingly.

Octreotide, the octapeptide analogue of somatostatin, is a powerful inhibitor of pancreatic exocrine secretion. A number of randomized prospective trials have examined the role of prophylactic, perioperative octreotide and its impact on outcome after pancreatic surgery. The results seemed to display an intercontinental difference: level 1 studies from Europe generally showed that perioperative somatostatin has a protective effect, whereas all the North American studies failed to demonstrate any benefit. A recent systematic review of this topic came to the conclusion that the prophylactic administration of somatostatin or its analogue does not uniformly reduce the incidence of pancreatic anastomotic leak, overall morbidity, or mortality after pancreatic resection (Li-Ling & Irving, 2001). Based on the findings of the trials conducted by our center (Büchler et al, 1992; Friess et al, 1995), we believe that some benefits may still be reaped from the use of prophylactic somatostatin in high-risk patients (soft pancreas, small duct), in whom a PD is performed. If the pancreas is deemed to be high risk by the surgeon, because of a soft consistency and a pancreatic duct size of less than 2 mm in diameter, somatostatin is applied intraoperatively (200 µg subcutaneously), followed by three daily doses of 200 µg octreotide for 5 days. Other indications for applying somatostatin are when segmental resections or enucleations must be performed in patients with a soft pancreas, as is frequently observed with cystic or endocrine lesions.

Preoperative Biliary Drainage

Patients with pancreatic cancer who have jaundice are also at risk for associated coagulopathy, malabsorption, malnutrition, and immune dysfunction. Using a multiple-variant regression analysis, we had previously determined that jaundice (bilirubin level >5.8 mg/dL) is a significant risk factor for postoperative hemorrhage (Martignoni et al, 2001). This calls into question the role of preoperative biliary drainage (PBD) in reducing surgical morbidity through the normalization of liver function and the correction of coagulopathy. Sewnath and colleagues (2002) found in a meta-analysis no difference in the overall death rate among patients who had PBD and those who had surgery without PBD. Instead, the overall complication rate was adversely and significantly affected by PBD, and the hospital stay was also prolonged, so they concluded that PBD carries no benefit. Recent studies also demonstrate that routine PBD should not be performed because of the high rate of procedure-associated complications (Mezhir et al, 2009; van der Gaag et al, 2010). We concur with the opinion that ERCP should only be undertaken with therapeutic intent and believe that PBD as a routine practice should be halted. We would refer a patient for PBD in the presence of cholangitis or other severe complication of jaundice that would preclude a safe, early resection. An absolute indication for PBD is jaundice that requires induction therapy prior to surgical extirpation.

Fast-Track Surgery

Studies on so-called fast-track GI surgery have shown that epidural analgesia, combined with an intensive and standardized regimen of early feeding and mobilization, can reduce hospital stay (Basse et al, 2002). Epidural analgesia has been found to have many attributes, including a shorter duration of postoperative ileus, attenuation of the stress response, fewer pulmonary complications, improved postoperative pain and mobility, and patient perception of a quicker recovery (Fotiadis et al, 2004). Thoracic epidural analgesia is of particular benefit to patients with a high risk of cardiac or pulmonary morbidity, and it can reduce hospital stay and costs in this subgroup of patients.

Attracted by the promising results published in the literature,and encouraged by the concept of fast-track surgery, we have since recommended thoracic epidural analgesia to all our patients who undergo pancreatic resection. We use a combination of local anesthetic and opioids; the former blocks sympathetic outflow, resulting in less cardiac stress, and at the same time promotes the return of bowel motility. The addition of opioids allows adequate analgesia with less local anesthetic, thereby reducing local anesthetic–related hypotension and motor deficits. Alternative methods of analgesia can be used when epidural analgesia is contraindicated or declined. We have also established postoperative pain teams to assist in patient care and monitoring and have realized some of the published benefits, including earlier and better tolerance of oral feeding, less postoperative pain, and earlier mobilization.

Technique

In thin patients, we prefer a midline incision beginning from the xiphoid process, but we perform a rooftop incision in patients with a short torso and a wide abdominal girth. After we have ruled out liver metastasis, the peritoneal surfaces are carefully inspected for metastatic deposits. Particular attention is paid to the pelvis for drop metastasis and to the root of the mesentery for evidence of tumor extension or matted nodes. A wide Kocher maneuver is then performed to assess the retroperitoneum and appraise the tumor and its relationship to the SMA (Fig. 62A.3). The mesenteric artery is inspected from the ligament of Treitz and dissected on its right hemicircumference to exclude any tumor infiltration of the artery. This technical approach, the so-called artery-first approach (Fig. 62A.4; Weitz et al, 2010), helps avoid R2 resections because any infiltration of the central arteries can be diagnosed early during exploration. We proceed with the resection only if we find no evidence of infiltration of the celiac trunk and SMA.

FIGURE 62A.3 Adequate kocherization to facilitate assessment of the retroperitoneum and the relation of the tumor to the vascular structures.

FIGURE 62A.4 Artery-first approach. The superior mesenteric artery is dissected, and infiltration of the artery is ruled out as one of the first steps of the operation.

Access to the lesser sac is achieved by division of the gastrocolic ligament. On the left side, we divide the gastrocolic ligament only as far as the most medial branch of the short gastric vesssls. This is to ensure an alternative venous egress for the splenic blood flow in the event of any venous resection of the SMV-PV trunk. Moving to the right, the hepatic flexure is mobilized caudally. Careful dissection in the avascular plane between the hepatic flexure and the duodenum and extension of the Kocher maneuver will allow the third part of the duodenum to be freed from the colonic mesentery. The gastroepiploic vein is then divided where it empties into the gastrocolic trunk, and the SMV is traced to the inferior margin of the pancreas. Two stay sutures are then placed at the inferior border of the pancreas to aid in the creation of the tunnel between the pancreatic neck and the SMV-PV trunk. Attention is now turned to the supraduodenal compartment.

Cholecystectomy is performed in a typical fashion. The cystic duct is traced to its origin from the common bile duct (CBD), which is transected just cephalic to this point. Extreme care must be taken to avoid any iatrogenic injury to the right hepatic artery, which usually runs posterior to the hepatic duct (see Chapter 1B; Skandalakis et al, 2004). In addition, it is important to assess for an aberrant right hepatic artery, present in 20% of patients, by palpation of the posterior aspect of the hepatoduodenal ligament. The cystic artery is isolated and closed with ligation.

The lymphadenectomy of the hepatoduodenal ligament is performed with great caution and completed along the common hepatic artery toward the celiac trunk. The distal end of the CBD and its adjoining fibrous fatty tissues are then dissected free from the rest of the hepatoduodenal ligament and retracted caudally. A small, noncrushing clamp is then applied to the proximal bile duct stump to prevent any further bile spillage for the rest of the operation. The proper hepatic artery is identified and looped then traced proximally toward the common hepatic artery. The right gastric artery (RGA) and the gastroduodenal artery (GDA) can be isolated during this dissection. Nodal tissues surrounding the proper hepatic artery and the common hepatic artery are excised. The RGA and GDA are then divided near the origin of the GDA. A potential pitfall here is the misidentification of a replacing common hepatic artery or even a replacing right hepatic artery as the GDA. A technique to avoid this mistake is to place a vascular clamp across the presumed GDA and check for pulsations at the porta hepatis before this vessel is divided.

If preservation of the pylorus is planned, which is our usual practice, the right gastric and right gastroepiploic arteries are divided and the duodenum is skeletonized 2 cm beyond the pylorus. The suprapancreatic portion of the PV will now be widely exposed, and two stay sutures are similarly placed on the superior border of the pancreas; these sutures at the superior and inferior pancreatic borders also serve to ligate the superior and inferior pancreatic vessels respectively, running longitudinally in the parenchyma. Hence bleeding from the cut edges is reduced after transection. A tunnel is cautiously created between the SMV-PV trunk posteriorly and the pancreatic neck anteriorly. A silicon drain is then insinuated into this tunnel to loop up the neck (Fig. 62A.5).

FIGURE 62A.5 Clearance of the surrounding fibrofatty tissues around the pancreatic stump for a distance of 2 cm allows adequate visualization of the pancreatic capsule. Cannulation of the pancreatic duct facilitates subsequent ductal sutures.

Because the medial margin is the most crucial with regard to tumor infiltration and the R status of the resection (Esposito et al, 2008), we now perform the uncinate process–first approach (Hackert et al, 2010). After having exposed and evaluated the SMA, the ligament of Treitz is mobilized, and the mesenteric branches to the fourth part of the duodenum are divided to allow it to be delivered into the inframesocolic compartment under the SMA. The mesenteric window is extended to the proximal jejunum, which is transected at a point that allows a tension-free loop of jejunum to be brought into the supramesocolic compartment in a retrocolic fashion through a window created on the right side of the middle colic pedicle. The dissection is performed along the SMA and SMV with bipolar cautery and clips.

If venous resection is required, it is reserved as the last step in the extirpative phase. The PV is then gently retracted medially to expose the underlying tissues, and any venous branches are divided. At the same time, the specimen is retracted to the right. The tissue and branch arteries arising from the SMA are serially clamped, divided, and clipped or stitch ligated. During this step, the specimen that comprises the pancreatic head, duodenum, and proximal jejunum is cupped in the surgeon’s nondominant hand, and the fingers continuously appraise the position of the SMA to avoid any injuries to it. The anterolateral aspect of the SMA is completely skeletonized of its investing tissues.

After complete transection of the pancreatic mesentery along the SMA and SMV, a scalpel is used to transect the neck of the pancreas. Bleeding vessels from the transected pancreatic ends are ligated using nonabsorbable monofilament polybutester sutures. We do not use any form of electrical cautery on the pancreas for hemostasis; we believe this to be an insecure form of hemostasis when used on the pancreatic parenchyma. We also suspect that this will be detrimental to the tissues and that it may compromise the subsequent anastomosis. The venous trunk is then examined for any tumor involvement on its posterolateral aspect, and margins are harvested from the proximal pancreatic stump and bile duct for margin analysis by frozen section.

At last, the duodenum is transected by using a linear stapler. The stomach is then retracted to the left upper quadrant of the abdomen, and the specimen is sent for histologic examination. The pancreatic stump is cleared over a distance of 2 cm from the cut end to provide a clear, all-around visualization of the pancreatic capsule and to allow a safe pancreatojejunostomy; this also facilitates the subsequent construction of the pancreatoenteric anastomosis. Next, hemostasis is ensured, and the operative field is washed with warm water before proceeding to the reconstructive phase.

Standard Pancreatoduodenectomy Versus Pylorus-Preserving Pancreatoduodenectomy

In addition to the known postoperative complications associated with PD, the performance of a distal gastrectomy in the standard procedure also brings with it the possible long-term morbidities of gastric dumping, marginal ulceration, and bile-reflux gastritis (Poon & Fan, 2001). First introduced by Kenneth Watson (1944), it was not popular until 1978, when Traverso and Longmire (1978) reintroduced it. They hypothesized that the preservation of the stomach and pyloric function would improve GI function and hence reduce the morbidities associated with a gastroenterostomy. Indeed, better long-term digestive function and quality of life have been found after pylorus-preserving pancreatoduodenectomy (PPPD) in some retrospective studies. To retain a functioning pylorus, the entire stomach, 2 cm of the first part of the duodenum, and the neurovascular supply to both are preserved. Following the division of the right gastric and right gastroepiploic arteries at their origin, the duodenum is skeletonized 2 cm distal to the pylorus (Farnell et al, 2001), taking care to preserve the nerve of Latarjet. The duodenal bulb is then transected.

A recent meta-analysis demonstrated that the classical Kausch-Whipple procedure and the pylorus-preserving Whipple procedure are comparable with regard to perioperative morbidity and mortality and long-term outcome (Diener et al, 2008b). Hence, in the light of the available evidence, PPPD should be designated the “standard” procedure for pancreatic and periampullary neoplasm because no study to date has demonstrated any advantage to the removal of the stomach and the proximal duodenum (Bell, 2004). Hemigastrectomy should be performed only when gastric or proximal duodenal invasion or grossly abnormal peripyloric nodes are evident.

Vascular Resection

Over 30 years ago, Fortner (1973) reasoned that a more radical resection should improve survival by improving tumor clearance, specifically that tumor adherence to the PV or SMV, often regarded as a criterion of unresectability, could be overcome by en bloc resection of the involved vessels. Randomized evaluation of portal and superior mesenteric vein resection in patients with tumors adherent to these vessels is difficult because considerable intersurgeon variation in the interpretation of adherence is likely. Thus no randomized trial evaluates this topic.

A recently published systematic review (Siriwardana & Siriwardena, 2006) evaluated 52 manuscripts with 6333 patients in whom pancreatic resection was performed for pancreatic cancer; 1646 of these patients (26%) underwent synchronous portal and superior mesenteric vein resection. The median number of resections per publication evaluated was 82; of those, 23 patients underwent portal and superior mesenteric vein resection. The proportion of portal and superior mesenteric vein resection per publication varies widely, ranging from 2% to 77%, which mirrors the various treatment approaches of pancreatic cancer in different institutions today.

Information on the operation was available in 39 studies with 1334 patients. The operations performed included PD (71%), total pancreatectomy (24%), subtotal pancreatectomy (3%), and distal pancreatectomy (2%). The median operation time was 513 minutes (range, 168 to 1740 minutes), median blood loss was 1750 mL (range, 300 to 26,000 mL), and the median time for PV occlusion was 20 minutes (range, 7 to 302 minutes). The perioperative mortality rate was 5.9%, varying between 0% and 33%, and postoperative morbidity rate was 42%, ranging from 9% to 78%. PV invasion was detected in 64% of all portal and superior mesenteric vein resection specimens, and the range varied widely between 3% and 86% in the different series, and 67% of all patients with portal and superior mesenteric vein resection had positive lymph nodes detected on histology. The long-term survival was 13 months (range, 1 to 109 months) after portal and superior mesenteric vein resection. The 1-, 3-, and 5-year overall survival rates of 1351 patients undergoing portal and superior mesenteric vein resection was 50%, 18%, and 8%, respectively.

The assessment of pooled data is critically influenced by the quality of the single reports. The reports included have been published over the last 25 years, and the standards of perioperative care, surgical technique, and adjuvant therapy have developed dramatically. Thus the wide variations of outcome parameters—perioperative death, operative time, blood loss—mirror the nonhomogenous collective included in this review. However, the long-term survival rate demonstrates that resection of the portal or superior mesenteric vein is a potentially curative operation. It seems that patients who need a portal or superior mesenteric vein resection for cure include a high percentage of patients with a positive nodal stage. Because adjuvant treatment has been proven to increase survival but has only recently been introduced as standard care for patients with pancreatic cancer, the survival rates of patients with portal or superior mesenteric vein resection today should be even better than those reported in this review (Neoptolemos et al, 2004; Oettle et al, 2007).

In addition, surgical expertise has improved over the past few decades, and the quality of high-volume centers—especially for technically demanding surgeries, such as pancreatic resection—has been proven (Birkmeyer et al, 2003; Büchler et al, 2003). Thus it is not surprising that perioperative morbidity and mortality rates reported for pancreatic resections with portal or superior mesenteric vein resection are identical to those without it (Bachellier et al, 2001; Carrere et al, 2006; Leach et al, 1998; Nakao et al, 2006; Riediger et al, 2006; Wagner et al, 2004).

These latest reports also demonstrate that extended resections with portal or superior mesenteric vein resection improve long-term survival. However, histologically confirmed tumor infiltration into the tunica media of the PV has a worse prognosis and a long-term survival of less than 2 years.

Tumor infiltration of the SMA, celiac trunk, or hepatic arteries are well-accepted criteria for unresectability of pancreatic adenocarcinoma. However, because adherence of tumors to the arteries does not automatically mean infiltration—in fact, the arterial wall is resected much longer than the venous wall—several published series of pancreatic resections included resection and reconstruction of visceral arteries (Hartwig et al, 2009a; Li et al, 2004; Sasson et al, 2002; Varadhachary et al, 2006). A recent review states that in 15% of pancreatic operations in which portal or mesenteric vein resection had to be performed, arteries were also resected. These included the common hepatic artery (50%), SMA (20%), celiac axis (10%), and other arteries (Siriwardena & Siriwardena, 2006).

In summary, resection of the portal and superior mesenteric vein should be performed when adherence or infiltration of the tumor is present, because perioperative morbidity, mortality, and long-term survival are identical to those patients in whom an R0 resection can be performed without PV resection. Moreover, the long-term survival of patients who have had pancreatic resection in combination with PV resection is far superior compared with those in whom only palliative surgical management was performed. In contrast, with arterial adherence of the pancreatic tumor, arteries can only rarely be resected in a curative way. Although resection of the SMA, celiac trunk, or hepatic arteries is technically feasible, no data suggest that this leads to any improvement of long-term outcome. Thus, resections of arteries during pancreatic resections are an experimental approach that cannot be recommended as a standard technique today.

Multivisceral Resections

The benefit of radical surgical resection of contiguously involved structures for locally advanced pancreatic cancer has been addressed in several reports (Hartwig et al, 2009; Sasson et al, 2002). In fact, about 35% of patients are seen initially with locally advanced pancreatic cancers with involvement of surrounding structures and organs. Although several reports in the past showed that morbidity of extended resection was increased and survival benefit limited, more recent publications demonstrate that en bloc resection of contiguously involved organs can be performed safely (Sasson et al, 2002). No difference was reported with regard to perioperative morbidity (35%) and mortality (3%) compared with standard resection. Although operation time is clearly longer because of the extended resection—which includes mesocolon, colon, adrenal glands, liver, and stomach—blood loss and hospital stay are not different from what is observed after the standard procedure. The main aim of the extended procedure must be an R0 resection because this is the most important predictor of long-term survival (Wagner et al, 2004). The survival rate after resection of mesocolon and colon, as well as stomach, is not significantly decreased compared with the standard procedure (Sasson et al, 2002). The 5-year survival rate was 16%, and the median survival of 26 months is much better than the 6 to 9 months reported for patients who are not resected at all.

Similarly, another study demonstrated that patients with pancreatic adenocarcinoma of the body or tail need extended resection for contiguous organ involvement more frequently because these tumors are generally detected later and have therefore reached a greater size than cancers of the pancreatic head by the time of diagnosis. The median survival in these patients was 15.9 months, and the survival at 5 and 10 years was 22% and 18%, respectively. In our series of more than 100 patients with multivisceral resections, we showed that perioperative mortality rate and long-term outcome are similar to those of standard resections (median survival, 20 months), but that the morbidity is increased. Thus these patients need careful postoperative observation in an experienced pancreatic center (Hartwig et al, 2009).

In summary, patients undergoing extended resection for adenocarcinoma of the pancreas have long-term survival rates similar to those undergoing standard resection. The main aim is to achieve a curative (R0) resection, and resected patients have improved survival compared with patients with advanced disease who are not resected or considered unresectable; thus extended pancreatic resection for contiguous organ involvement is justified.

Lymphadenectomy

The rationale for extended lymphadenectomy is that lymph node studies have confirmed positive lymph nodes may be found outside the confines of the standard dissection (Cubilla et al, 1978). Even for small cancers, lymph nodes of the paraaortic region, between the celiac trunk and the origin of the inferior mesenteric artery, frequently harbor metastases; it has been suggested that these should be dissected en bloc during radical resections (Nagakawa et al, 1993). These revelations initiated a movement of extended lymphadenectomy among Japanese surgeons. A few studies reported similar surgical morbidity and mortality rates, but improved survival results were seen with extended surgery compared with the standard PDs (Ishikawa et al, 1993). However, these were all retrospective, nonrandomized studies.

What constitutes extended lymphadenectomy is still widely debated; this is epitomized by the multitude of terminologies used in the literature. In an attempt to harmonize this subject, a consensus conference on the surgical treatment of pancreatic cancer took place in Castelfranco Veneto, Italy (Pedrazzoli et al, 1999). This consensus provided a standardized definition of the different extent of lymphadenectomy that used the Japanese Pancreas Society rules for the study of pancreatic cancer to define the lymph node stations to be removed for the various procedures (Japan Pancreas Society, 1993). The three stages of radicality for PD have been named standard, radical, and extended radical, depending on the nodal stations removed. For cancers of the body or tail, according to the extent of lymphadenectomy, two different procedures are identified: standard and radical. It is hoped that this standardization of surgical procedures will facilitate comparison of results between different centers.

Between 1988 and 1999, eight retrospective studies were published on the extent of lymphadenectomy for PD. The data on survival are largely heterogeneous, but neither morbidity nor mortality seems to be altered. Altogether, two studies reported on morbidity rates, whereas five studies reported perioperative mortality rates. Especially in Japan, extended lymphadenectomy was widely used in the late 1980s and 1990s because of the initial reports of Ishikawa and colleagues (1988) and Manabe and colleagues (1989), who advocated an extension of the PD. In these studies, survival rates were significantly increased in the extended lymphadenectomy groups. Interestingly, some long-term survivors were seen in the extended lymphadenectomy groups, but none (Manabe et al, 1989) or very few (Ishikawa et al, 1988) of the patients in the standard lymphadenectomy arm survived for more than 5 years.

In terms of operative techniques, total pancreatectomies were used in many patients, which may have also altered the outcome. Interestingly, the largest retrospective studies (Hirata et al, 1997; Mukaiya et al, 1998), which together included 1330 patients, did not reproduce these results. The 1- and 3-year survival rates were comparable in the standard lymphadenectomy and extended lymphadenectomy groups, but data on morbidity and mortality were not provided. A study by Kawarada and colleagues was the only one to reproduce the results of Manabe and Ishikawa. However, patients included in the standard lymphadenectomy arm had been operated on between 1976 and 1981, whereas extended lymphadenectomy was performed between 1981 and 1993. Because the perioperative outcomes were significantly improved in the late 1980s, the survival benefit shown for extended lymphadenectomy may be due solely to generally advanced perioperative management. Hence, it cannot be concluded from the retrospective studies that extended lymphadenectomy provides a survival benefit.

Six prospective, nonrandomized studies have been published on this topic. A prospectively conducted study from Henne-Bruns and colleagues (1998), in which the decision to perform a standard or extended lymphadenectomy was made based on the clinical status of the patient, revealed no significant differences in mortality rate and long-term survival but did show a trend toward a better outcome in the standard lymphadenectomy group. This may be due to a higher percentage of patients with advanced-stage tumors in the extended lymphadenectomy group. Furthermore, extended lymphadenectomy was associated with an increased rate of severe diarrhea in the early postoperative phase. Gazzaniga and colleagues (2001) showed that morbidity, mortality, and long-term survival were unaltered between standard and extended lymphadenectomy. Interestingly, a third group of patients received extended lymphadenectomy and additional adjuvant chemoradiation and fared significantly better, with a 5-year survival rate of 17.6%. However, this promising concept of multimodal therapy was not confirmed by randomized trials. Another retrospective study from Italy showed a tendency toward increased survival rates with an extended lymphadenectomy, but morbidity was comparable (Iacono et al, 2002). Studies from both Popiela and colleagues (2002) and Capussotti and colleagues (2003) confirmed the results of the previous reports with unchanged morbidity, mortality, and survival rates. However, for node-negative cancer, extended lymphadenectomy significantly improved the 5-year survival rates (Popiela et al, 2002), and a slight improvement in short-term survival was also seen for patients in the extended lymphadenectomy group (Capussotti et al, 2003). Overall, the number of resected lymph nodes was significantly higher in the extended lymphadenectomy groups.

Four randomized controlled trials have been conducted so far and were summarized in a meta-analysis (Farnell et al, 2005; Michalski et al, 2007; Nimura et al, 2004; Pedrazzoli et al, 1998; Yeo et al, 2002). Interestingly, these four level I studies are from centers on three different continents. The first study by Pedrazzoli and colleagues did not reveal any differences with regard to morbidity, mortality, and survival between the standard and extended lymphadenectomy groups. Standard lymphadenectomy included the dissection of lymph nodes at the anterior and posterior pancreatoduodenal side, at the pylorus, bile duct, and superior and inferior part of the pancreatic head and body. For the extended procedure, additional lymph nodes were dissected within the complete circumference at the aorta, the inferior and superior mesenteric arteries, and the celiac trunk. Interestingly, a retrospective subgroup analysis revealed a significantly prolonged survival for node-positive patients who received an extended lymphadenectomy. This statement has since been intensively discussed because of the small difference in the number of resected lymph nodes between the standard and extended lymphadenectomy groups (13.3 vs. 19.8).

The largest study on this subject enrolled 294 patients; it was published in 2002 and updated in 2005 by Yeo and colleagues and was preceded by a preliminary evaluation in 1999. The authors randomized 294 patients into standard and extended lymphadenectomy arms after intraoperative frozen section with tumor-negative margins. However, the study encompasses the different entities of periampullary malignancies, and only 57% of the tumors were of pancreatic origin. Thus conclusions with regard to survival have to be critically reviewed because ampullary and distal bile duct tumors were also included. A pylorus-preserving PD was performed in 86% of the patients in the standard lymphadenectomy group; in the extended lymphadenectomy arm, the classic Whipple operation was mainly used. Mortality and survival rates were unaltered, whereas a significant increase in morbidity was reported in the extended lymphadenectomy group (29% vs. 43%; P < .01), which was mainly due to higher rates of delayed gastric emptying and pancreatic fistula.

Considering the large number of patients and the detailed analysis of many variables, the conclusion that extended lymphadenectomy does not improve survival but may increase morbidity is to be supported. These data and conclusions are underlined by their follow-up publication and a study on quality of life (QOL) of the same patient group (Riall et al, 2005). Additional studies from the Mayo Clinic by Farnell and colleagues (2005) and from Japan by Nimura and colleagues (2004) supported those results. The Mayo Clinic trial included 79 patients randomized in two groups (lymphadenectomy: standard [n = 40] vs. extended [n = 39]). When performing an extended lymphadenectomy, the number of lymph nodes resected was greater (36 vs. 15), but morbidity and mortality rates were comparable. No significant differences were observed for the 1-, 3-, and 5-year survival rates. However, at 4 months postoperatively, diarrhea was significantly increased, and bowel control and body appearance were worse in the extended lymphadenectomy group. The multicenter trial from Nimura and colleagues revealed similar results with comparable mortality and survival rates but increased morbidity, which was due to severe diarrhea in 48% of the patients (Nimura et al, 2004). Of interest, the number of lymph nodes resected varies considerably between the studies. For examples, Nimura and colleagues resected about twice as many lymph nodes in the extended lymphadenectomy group as did Pedrazzoli and colleagues, which may reflect different operative techniques as well as differences in pathologic analysis.

In summary, none of the randomized, controlled trials (RCTs), except for a retrospective subgroup analysis within the study of Pedrazzoli and colleagues, showed any survival benefit for extended lymphadenectomy, whereas overall morbidity with diarrhea and delayed gastric emptying as particular complications tended to occur more frequently. Interestingly, both Yeo and colleagues and Farnell and colleagues particularly report a reduction in QOL in the extended lymphadenectomy group, primarily because of high rates of postoperative diarrhea with subsequent development of malnutrition (Michalski et al, 2007). Thus mortality and morbidity may be at best unaltered for the extended radical operation compared with the standard procedure. The randomized studies are altogether underpowered and lack standardization with regard to the actual extent of the performed lymphadenectomy. As for the potential survival benefit, further adequately powered randomized studies will need to be impractically large, thus precluding their realization. Therefore, the current standard of surgery for pancreatic head tumors remains PPPD without extended lymphadenectomy.

Reconstruction After Pancreatic Resection and Management of the Pancreatic Remnant

A pancreatic leak is the major cause of procedure-related death. Simply occluding the duct has been shown to result in higher fistula rates in addition to increasing the risk of pancreatic exocrine and endocrine insufficiency (Tran et al, 2002). Hence, drainage of the pancreatic remnant to the GI tract remains a crucial step, but it runs the risk of anastomotic breakdown. It is thus not surprising that the pancreatoenteric anastomosis has fascinated surgeons, motivating them to search for a more reliable technique to avoid this dreaded complication. Many techniques have been described, and the literature will continue to report novel techniques that promise to be even safer. However, more so than the choice of the variant used, successful management of a pancreatic anastomosis is most dependent on the surgeon’s concentration on meticulous execution of the chosen technique (Trede et al, 2001). For as long as the basic tenets of a safe anastomosis are met—namely, careful handling of the pancreatic tissues, a tension-free adaptation, good perfusion, and no distal obstruction—any pancreatoenteric anastomotic technique can have a good outcome.

One of the most commonly used techniques is a pancreatojejunal anastomosis, which can be performed by invaginating the transected pancreas into the end of the jejunum, the so-called dunking procedure. Another variant is to anastomose the pancreatic duct directly to a proper opening in the jejunum, the so-called duct-to-mucosa technique. The technique of pancreatojejunal anastomosis—whether end-to-side or end-to-end, duct-to-mucosa or dunking—does not seem to significantly influence the anastomotic leak rate. Another strategy is to anastomose the pancreatic stump to the stomach, and proponents of pancreatogastrostomy (PG) have cited various reasons to choose this method (Zenilman, 2000); it is easier to perform given the close proximity of the stomach to the pancreas, and the anastomosis is less prone to ischemia because of the rich gastric perfusion. Because the exocrine enzymes enter an acidic environment, it was once thought that the leak rate would be lower because the enzymes do not get activated, but this idea has since been debunked. In a prospective randomized trial comparing pancreatojejunostomy (PJ) to PG, the leak rates were not significantly different (PJ, 11%; PG, 12%) (Yeo et al, 1995).

The authors’ institution uses a standardized technique of PJ performed in an end-to-side fashion with a retrocolic jejunal limb (Z’graggen et al, 2002). The anastomosis is performed in two layers with duct-to-mucosa adaptation using monofilament absorbable polydioxanon sutures (PDS 5/0) with an atraumatic JRB-1 (5/8) needle. The first step is the placement of the ductal sutures, first anteriorly from the outside in, then posteriorly from the inside out (Fig. 62A.6). We endeavor to place at least three ductal sutures on each side. This is generally no problem, because the duct is dilated in cases of pancreatic head tumors. Only in patients with very small ducts that cannot be cannulated with a small ductal probe, the duct sutures should be omitted to avoid obstruction of the duct.

FIGURE 62A.6 Three anterior and three posterior ductal sutures.

Because the pancreatic stump is dissected over 2 cm toward the left, good visualization of the posterior capsule can be achieved. A row of interrupted sutures is then placed through the dorsal capsule, with the needle entering the pancreatic tissues about 1 cm from the cut end of the stump and taking an adequate bite of the parenchyma. The needle is then driven through the seromuscular layer of the jejunal limb about 1 to 2 mm from the mesenteric border. The intersuture distance is about 4 to 6 mm (Fig. 62A.7). The jejunum is carefully adapted to the pancreas following meticulous and gentle tying of the knots. A longitudinal enterotomy is then performed about 5 mm from the posterior row of sutures (Fig. 62A.8) with the width of the opening tailored to the size of the pancreatic stump.

FIGURE 62A.7 The first layer of the posterior suture row performed in an interrupted fashion.

FIGURE 62A.8 Following apposition of the pancreas to the jejunal limb via the posterior serocapsular suture row, a longitudinal enterotomy is planned, tailored to the width of the pancreatic stump.

The posterior inner row of sutures is then constructed, first catching the pancreas in an inside-out fashion, ensuring an adequate bite of the parenchyma, and exiting just at the cut edge of the remnant pancreas. The suture is then driven through the posterior edge of the enterotomy in an outside-in fashion, ensuring a full-thickness bite. The previously placed ductal sutures are integrated in order, and each suture is clipped with an artery clamp (Fig. 62A.9). All knots are tied after completion of the suture row.

FIGURE 62A.9 The completed second layer of the posterior suture row demonstrates the integration of the previously placed ductal sutures into the suture row.

This is followed by the anterior inner row of sutures, which is driven through the pancreas in an outside-in manner, entering just at the anterior cut edge of the stump and ensuring an adequate bite of the pancreatic parenchyma. The suture is then placed through the anterior edge of the enterotomy in an inside-out fashion with a full-thickness bite (Fig. 62A.10). All knots are, again, only tied following completion of the suture row.

FIGURE 62A.10 Placement of the first layer of the anterior suture row.

The final step is the second row of anterior sutures, the aim of which is to cover the inner suture line by imbricating the jejunal serosa over it (Fig. 62A.11). The suture first catches the anterior capsule of the pancreas close to the anterior inner row. A seromuscular bite is then taken of the jejunum at a point that will allow a fold of the jejunal wall to cover the inner row of sutures in a tension-free manner (Fig. 62A.12).

FIGURE 62A.11 The second layer of the anterior suture row, which serves to imbricate the first layer.

FIGURE 62A.12 Completion of the pancreatojejunostomy.

The end-to side technique allows the adaptation of the jejunal opening to the specific requirements of the pancreatic remnant. Separate duct-to-mucosa adaptation also keeps the duct orifice open, thereby ensuring the unobstructed flow of pancreatic secretions through the anastomosis. Consequently we see no need for the placement of any pancreatic stents. This technique was used in 93% of patients undergoing pancreatic head resection, with an overall leak rate of 3.2% (Büchler et al, 2003).

Bilioenteric Anastomosis

Unlike the pancreatoenteric anastomosis, fewer variations are used for a bilioenteric anastomosis, which is usually constructed in an end-to-side fashion with a single layer of sutures using monofilament absorbable sutures (PDS 5/0) with a C1 needle. This needle, with its smaller rounding radius, enables easier handling, especially within the confines of the bile duct, particularly when the caliber of the duct is small.

We position this anastomosis about 10 to 15 cm downstream from the PJ. The cut edge of the bile duct is freshened to obtain well-perfused healthy tissues, after which a small enterotomy is made at the antimesenteric border of the jejunal loop, with removal of a small cuff of mucosa. The protruding mucosa is then fixed at four quadrants of the enterotomy using 6/0 PDS sutures (Fig. 62A.13). These mucosa-fixation sutures ensure a full-thickness bite in the subsequent suture placement and thus guarantee a secured duct-to-mucosa coadaptation. The posterior suture layer is first constructed, with each individual suture held in clamps, until the entire suture row is completed and before knotting begins (Fig. 62A.14). This is then followed by the anterior suture row (Fig. 62A.15), and the final step is closure of the mesocolic opening around the jejunal limb.

FIGURE 62A.13 Jejunal mucosal fixation stitches.

FIGURE 62A.14 Posterior suture row of the biliojejunal anastomosis in interrupted fashion.

FIGURE 62A.15 Anterior suture row of the choledochojejunostomy.

Reconstitution of Gastrointestinal Continuity

Depending on whether a distal gastrectomy or PPPD is performed, the reconstruction ends either with a gastrojejunostomy or a duodenojejunostomy, respectively. Some concern has surrounded the question of whether an intact pylorus will lead to higher delayed gastric emptying (DGE) rates postoperatively following a PPPD; however, this controversy seems to have been finally laid to rest following the results of several controlled trials that to which we have previously alluded. Therefore, based on evidence available in 2004, it is apparent that pylorus preservation does not seem to increase the frequency of DGE (Horstmann et al, 2004).

The technique used to reconstitute GI continuity may have an impact on the incidence of DGE, however. One group speculated that a retrocolic reconstruction predisposes the jejunal limb to venous congestion and bowel edema, which can consequently retard recovery of jejunal peristalsis at the duodenojejunostomy (Park et al, 2003). Postoperative gastroparesis may also lead to temporary gastric distension, which can lead to angulation of the anastomosis because it lies relatively fixed through its retrocolic position (Horstmann et al, 2004). Additionally, the close proximity of the duodenojejunostomy to the PJ also predisposes the occurrence of DGE in the event of a small PJ leak or transient postoperative remnant pancreatitis. Opponents to this theory have suggested that the real culprit is an antecolic reconstruction (Riediger et al, 2003) that predisposes the relatively fixed stomach to angulation or torsion. The risk of DGE caused by local inflammation is reduced by placing the duodenojejunstomy in the infracolic compartment through a separate mesenteric window, away from the pancreatic and biliary anastomoses, which lie in the supracolic compartment.

Our unit’s philosophy is more in line with the first group, substantiated by the results of a retrospective review conducted in our center, in which we compared the outcomes following retrocolic placement of the duodenojejunostomy with those following an antecolic placement of the anastomosis. We found that the incidence of DGE was lower after using the second surgical strategy. After PPPD, we now routinely perform an antecolic end-to-side duodenojejunostomy about 50 cm downstream from the hepaticojejunostomy, using the transverse colon to splint the duodenojejunostomy away from the PJ. The anastomosis is constructed in two layers using monofilament absorbable sutures in a continuous fashion (PDS 4/0 or 5/0) (Fig. 62A.16).

FIGURE 62A.16 Completion of the reconstruction following the pylorus-preserving pancreatoduodenectomy.

Abdominal Drains and Nasogastric Tubes

Intraperitoneal drains have been placed in relation to the biliary and pancreatic anastomoses with the intention of controlling leakage of blood and biliary, lymphatic, or pancreatic secretions. Such a practice has been prophylactic in nature, and the reason for it stems more from habit than evidence. But this has been challenged by a randomized trial that addressed the value of drains following pancreatic resection and found that placement of drains did not translate into a reduction in surgical morbidity (Conlon et al, 2001). Rather, a significantly higher proportion of patients randomized to the drain group developed intraperitoneal sepsis, fluid collection, or fistula, although this trial was not sufficiently powered to determine whether the drain was the cause of these intraabdominal complications. Consequently, based on this trial alone, our unit has not found sufficient evidence for us to abandon this practice.

Our hesitancy to embrace a no-drain policy is further vindicated by the findings of a recently published meta-analysis of prophylactic drainage after GI surgery (Petrowsky et al, 2004). This evidence-based review found that drains do not reduce complications following hepatic, colorectal, or rectal resection with primary anastomosis, and drains should not be used following such procedures. With regard to many other GI procedures, including pancreatic resections, the jury is still out, and we have concluded that more well-designed trials are needed to clarify the role of prophylactic drainage. For our participation in a multicenter randomized (PANDRA) trial evaluating this issue, we routinely place two passive drainage tubes (Easy-flow; Dahlhausen & Company, Cologne, Germany): one below the diaphragm and the other in close proximity to the hepaticojejunostomy and the PJ.

We also routinely place nasogastric (NG) suction tubes for our patients during the operation and anesthesia, which are immediately removed at extubation. A meta-analysis on the need for NG decompression has quashed the long-held belief that routine NG decompression decreases the risk of postoperative nausea, vomiting, aspiration, wound dehiscence, and anastomotic leak (Cheatham et al, 1995). Instead, it was found that fever, atelectasis, and pneumonia were significantly less common, and days to first oral intake were significantly fewer, in patients managed without NG tubes. Although patients may develop abdominal distension or vomiting without an NG tube, this does not necessarily translate into an increase in complications or length of stay.

Technique

Following a thorough examination of the peritoneal cavity, the gastrocolic ligament is divided sufficiently to allow full visualization of the pancreas to its tail and the hilum of the spleen. Extreme care is taken here to preserve the short gastric vessels and the gastroepiploic arcade. The point of division is selected in the proximal normal pancreas, and the overlying peritoneum is divided along the superior and inferior borders of the pancreas. Stay sutures are then placed on either side of the planned transection line on both the superior and inferior borders. A plane is then developed posterior to the pancreas by blunt dissection, aided by upward tension on the stay sutures, until the splenic vein (SV) is identified. The splenic vein is situated posterior to the pancreas and lies fairly deep in the pancreatic parenchyma distally. Therefore, if problems in identifying the SV are encountered, one suggestion is to identify the SV-PV junction under the pancreatic neck, which is technically easier, and subsequently follow the SV toward the spleen. The splenic artery can usually be located along the superior border of the pancreas. The artery and the vein are then individually encircled with vascular loops. Following vascular control, our preference is early division of the pancreas.

Management of the Pancreatic Remnant

Pancreatic fistula remains the most common surgery-related complication after distal pancreatectomy (Benoist et al, 1999; Ferrone et al, 2008). Many techniques have been described in the literature with regard to the management of the residual transected pancreatic parenchyma and the divided pancreatic duct to lower the risk of leak. These include direct duct ligation, enteric drainage, prolamine injection, sealing with fibrin glue, and transection with linear stapling devices. The plethora of surgical strategies described in the medical literature simply underlies the fact that no one technique has been convincingly shown to reduce the incidence of pancreatic leak consistently.

When the pancreas has a soft texture and is not too thick, we transect it with a linear stapling device (Fig. 62A.17). This device gives a staple line consisting of a triple row of closely placed staples. In all other cases, the pancreas is dissected by a knife, and the resection margin is closed with single PDS stitches; the pancreatic duct should be identified and closed by separate sutures. A meta-analysis comparing the safety of hand-sewn sutures versus staple closure of the pancreatic remnant after pancreatic left resection did not show a significant difference with regard to morbidity, including postoperative fistula (Knaebel et al, 2005). Results of a European multicenter randomized trial comparing the two techniques over the last several years will be available shortly (Diener et al, 2008a).

FIGURE 62A.17 Staple transection of the pancreas.

Alternatives to these standard techniques are to cover the pancreatic remnant with omentum, small bowel, or the falsiform ligament and to perform a PJ to the left remnant. However, none of these techniques has been evaluated in larger randomized trials.

Splenic Preservation

Conventionally, the spleen has been removed en bloc with distal pancreatectomy. This was believed to be necessary because of the close relationship of the splenic artery and vein to the body of the pancreas; splenic preservation might compromise the oncologic resection (Shoup et al, 2002). Furthermore, this was thought to be technically simpler. Splenic preservation can be performed either with the division of the splenic artery and vein or with preservation of the entire length of both vessels. The paramount prerequisite in the former surgical approach is preservation of the gastrosplenic vessels to ensure splenic perfusion and allow venous drainage. In the latter technique, the splenic artery and vein must be uninvolved by tumor. Because of the important immune function of the spleen, it is not surprising that the incidence of infectious complications that require intervention is significantly higher in the splenectomy group (Shoup et al, 2002). More importantly, Schwarz and colleagues (1999) found that in their patients undergoing resection of pancreatic adenocarcinoma with curative intent, the median actuarial survival was 12.2 months with splenectomy versus 17.8 months without splenectomy. Such an adverse effect of splenectomy on survival was also evident when it accompanied surgical resection for gastric and colon cancer. Based on the available evidence, we concur with Conlon and colleagues (2001) in concluding that splenic preservation should be attempted in patients with low-grade malignancies or other more indolent tumors of the pancreatic body/tail, but we feel that splenic preservation may compromise oncologic clearance if performed for pancreatic adenocarcinoma.

Our preference is to attempt to preserve the splenic artery and splenic vein to reduce the risk of splenic necrosis and abscess formation; because splenic perfusion based on the short gastric vessels alone has been shown to be potentially inadequate (Gurleyik et al, 2000). Following division of the pancreas, the distal stump is elevated and rotated gently to the left. Multiple small, short branches of the splenic artery and vein are then identified and clipped serially with hemostatic clips and cut. The dissection proceeds from the proximal to distal pancreas, until the tail of the pancreas can be separated from the splenic hilum (Fig. 62A.18). This separation of the splenic vein from the pancreas can only proceed in the direction of the spleen because the vein has already divided into small vessels that can easily be injured at the level of the hilum. On the other hand, isolation of the pancreas from the splenic artery is described to be easier when performed from the hilum toward the pancreatic head (Jovine et al, 2004). This is also a less tedious task because there are fewer branches from the splenic artery, and they are all on one side. Any vessel injuries can be repaired by suturing with monofilament 5/0 vascular sutures. If attempts at hemostasis of such vascular injuries are unsuccessful, we divide the splenic pedicle. We avoid mobilizing the spleen medially into the operative field, as this step is unnecessary and carries the risk of iatrogenic splenic injury.

FIGURE 62A.18 Spleen-preserving distal pancreatectomy.

Technique

The lesser sac is entered, and the anterior aspect of the pancreas is widely exposed by dividing the adhesions between the posterior surface of the stomach and the pancreas. This is followed by intraoperative pancreatic ultrasonography to better delineate the lesion and to exclude concomitant lesions that may necessitate a change of plan. The pancreas is then elevated off the SMV-PV trunk. Stay sutures are placed in the superior and inferior pancreatic margins to indicate the proximal and distal limits of division and to aid in the subsequent dissection of the pancreas from the splenic vein. The pancreas is then divided proximally with a vascular stapler at least 1 cm from the lesion. The distal stump is then gently retracted toward the left to allow the cautious and tedious process of freeing the splenic vein from the posterior surface, clipping and cutting all the fine venous tributaries lying between the splenic vein and the pancreas. The pancreas is then divided distal to the lesion with a blade (Fig. 62A.19), and a specimen is sent for analysis and margin assessment.

FIGURE 62A.19 Excision of a segment of the pancreas with identification of the underlying splenic vein prevents iatrogenic injury to the vein.

Our usual technique of reconstruction follows that described in most other series and consists of the closure of the proximal stump and a Roux-en-Y PJ for the distal stump as described. The duct in the proximal pancreatic stump is closed by PDS single stiches, and the resection margin is covered by a serosal patch by apposing the serosa of the jejunal limb to the stump and taking seromuscular bites of the jejunum and the pancreatic capsule (Figs. 62A.20 to 62A.24).

FIGURE 62A.20 Following a two-layer pancreatojejunostomy to the distal stump, a serosal patch is performed on the stapled end of the proximal stump.

FIGURE 62A.21 Segmental chronic pancreatitis secondary to trauma with stenosis of the pancreatic duct. The patient presented with recurrent pancreatitis.

FIGURE 62A.22 Intraoperative finding of segmental inflammation secondary to trauma. The gastroduodenal artery (a.) and the superior mesenteric vein are demonstrated. The arrows indicate the margin of the inflamed area.

FIGURE 62A.23 Intraoperative finding after the segmental resection has been performed. The resection margin on the right has been closed with monofilament absorbable polydioxanon single stitches; the pancreatic duct is to the left.

FIGURE 62A.24 After reconstruction, the end-to-side pancreatojejunostomy and the coverage of the resection margin at the pancreatic head with a serosal layer with the same jejunal loop can be seen.

However, in cases in which stenosis or obstruction of the proximal pancreatic duct is suspected, we prefer to perform jejunal anastomosis for both the proximal and distal stumps. Such proximal ductal abnormalities can be rather prevalent in patients with chronic pancreatitis; therefore, by anastomosing the proximal stump to the jejunum, an alternative outflow route is created for the pancreatic secretions. We also prefer a double anastomosis when we have doubts about obtaining a secured stapled closure, such as when the proximal pancreas is found to be large and bulky.

A jejunal Roux limb is created by dividing the proximal jejunum 15 to 20 cm past the ligament of Treitz. This is then brought to the supracolic compartment in a retrocolic fashion, through a mesenteric window created on the left side of the middle colic vessels. The distal pancreatojejunal anastomosis is constructed first in the same two-layer fashion as previously described (see Management of the Pancreatic Remnant). This is then followed by the proximal anastomosis using the same jejunal Roux limb. The reconstruction is completed by an end-to-side jejunojejunostomy, and drains are placed.

Technique

In cases of main duct IPMN, an en bloc total pancreatectomy without transaction of the pancreas may be necessary. In this case, the mobilization of the pancreatic head and the transection of the jejunum and duodenum are performed first. The pancreatic tail is mobilized together with the spleen, and the dissection of the pancreatic mesentery along the PV and the SMA is the last step of resection (Werner & Büchler, 2010).

In cases with a positive pancreatic resection margin, we would already have resected the pancreatic head along with the adjoining duodenal loop and the distal CBD. The distal duodenum is then delivered into the infracolic compartment after taking down the ligament of Treitz, and the proximal jejunum is transected to complete the resection of the duodenum. The pylorus is preserved in the same manner as in a PPPD. If a TP is necessary because of cancer involvement of the distal stump, we would resect the pancreas together with the spleen to ensure radicality. The short gastric vessels are divided, the splenic artery is divided at its origin, and the splenic vein is divided at its confluence with the SMV-PV trunk with a vascular stapler or suture closure. The distal pancreatic stump and underlying splenic vein are retracted to the left, and the specimen is dissected off the retroperitoneum toward the spleen. The inferior mesenteric vein is ligated and divided, and the lienorenal ligament is divided.

For more low-grade tumors, such as IPMNs, we would attempt to preserve the spleen. In this case, we would preserve all the short gastric vessels in the event that the splenic vessels need to be divided. This is followed by the tedious process of ligating or clipping and cutting the multiple branches that connect the splenic vein and the splenic artery to the pancreas in an attempt to preserve both splenic artery and vein. Reconstruction after TP consists of an interrupted end-to-side, single-layer hepaticojejunal anastomosis with a retrocolic jejunal limb. An antecolic end-to-side duodenojejunostomy is then constructed about 50 to 60 cm downstream from the bilioenteric anastomosis (Kulu et al, 2009).

Technique

Because the ampulla is absent in nearly two thirds of patients (Sarmiento et al, 2001), it would be more often correct anatomically to use the term TDE of the papilla of Vater. The abdomen is entered through a midline incision in case a conversion to a more radical procedure is required. A generous Kocher maneuver is then performed to completely mobilize the second part of the duodenum, and three to four stay sutures are placed alongside the antimesenteric border on its anterior aspect to help elevate the duodenum toward the abdominal wound to facilitate the duodenotomy. A 6- to 8-cm long longitudinal duodenotomy is then made in a lazy-S fashion along the antimesenteric border opposite the pancreas; the incision is kept open with stay sutures (Fig. 62A.25). Circumferential stay sutures, using monofilament absorbable PDS (5/0), are placed about 1 cm beyond the gross border of the lesion and are kept long. Excision of the lesion is undertaken with electrocautery to minimize bleeding, along a resection margin just inside the circle formed by the circumferential stay sutures. A heavy stay suture is placed through the tumor mass to allow the tumor and the papillary region to be drawn upward. The dissection is carried around the tumor area in a stepwise fashion, and the incision must allow a full-thickness excision of the tumor that includes the biliopancreatic ductal junction. Ductal resection can be safely performed up to 1 cm (Nikfarjam et al, 2001); this is to safeguard against any possible tumor infiltrations of the bile and pancreatic ducts. As each duct is identified, stay sutures are placed through the ductal wall following complete transection (Fig. 62A.26). Following complete excision, the specimen is carefully oriented for the pathologist to facilitate frozen-section examination. If clear margins are reported, we would proceed with the reconstruction, which begins with the suturing together of the pancreatic and common bile ducts in a common septum (Fig. 62A.27). This is followed by the reimplantation of this new common channel back into the duodenal wall with 5/0 monofilament absorbable sutures to approximate the duodenal wall to the joined pancreatic and bile duct openings (Fig. 62A.28). Patency of the ducts is ensured before closure of the duodenotomy. The duodenal opening is then closed in two layers (Fig. 62A.29), and drains are routinely placed in the vicinity of the duodenotomy.

FIGURE 62A.25 A generous duodenotomy and the use of stay sutures allow good access to the papilla.

FIGURE 62A.26 Ductal sutures are placed once each duct is identified.

FIGURE 62A.27 Apposition of the pancreatic duct and bile duct to form a common septum.

FIGURE 62A.28 Reimplantation of the ducts to the duodenal mucosa.

FIGURE 62A.29 Closure of the lazy S–shaped duodenotomy in two layers.

Postoperative Management

Following operation, we routinely monitor our patients in the intensive care unit or the high-dependency unit for the first 24 hours. The NG tube is usually removed at extubation or, at the latest, within the first 12 hours after extubation, when the patient has regained a reasonable level of consciousness and has an intact cough reflex. Intravenous antiemetics are given prophylactically. The abdominal drains are usually removed within the first 3 days after operation, based on the volume and the amylase/lipase content (Bassi et al, 2005; Molinari et al, 2007). However, if the volume is excessive (>700 mL/day), the drains should be kept in place. If the fluid amylase level is three times the serum amylase level or higher, the drains are kept in and the patient is treated as for a pancreatic leak.

The issue of postoperative nutritional support has been a popular topic for research, which seems justified, since many patients with pancreatic disease present with significant weight loss. Two issues are critical with regard to the provision of nutrients postoperatively: timing and method of administration. Lewis and colleagues challenged the long-held practice of “nil by mouth” after GI surgery and conducted a systematic review of the topic and a meta-analysis of controlled trials. They showed that early feeding within 24 hours after surgery, whether by mouth or directly into the small bowel, can significantly reduce both the risk of any type of infection and the mean length of hospital stay. Risk reductions were also seen for anastomotic dehiscence, wound infection, pneumonia, intraabdominal abscess, and mortality, but these failed to reach significance.

On the question of the ideal route of administration, investigators at MSKCC conducted a prospective randomized trial of early enteral feeding (EEN) with an immune-enhancing formula in patients with upper GI malignancies. They found that EEN via a jejunostomy tube did not provide any benefit over postoperative crystalloid support in well-nourished patients (Heslin et al, 1997). A more recent publication echoed similar findings. Braga and colleagues (2001) observed that EEN did not improve outcome compared with total parenteral nutrition (TPN) in the overall population; however, they did find that in a subgroup of malnourished patients, EEN was associated with a significantly shorter length of stay. This was probably secondary to a lower overall complication rate, particularly the rate of infectious complications—noted in the EEN group compared with the TPN group. The prevalence of glucose and electrolyte disturbances was also lower during EEN than during TPN. In addition, researchers were able to demonstrate better gut oxygenation with EEN, not to mention significant cost savings compared with TPN. This makes EEN the ideal method for nutritional support, albeit in a selective manner.

EEN is not without risks, however. We previously demonstrated a significantly higher incidence of DGE in the enteral nutrition (EN) group compared to the no-EN group (Welsch et al, 2010). Furthermore, the use of a feeding jejunostomy tube is itself subject to the risk of adverse events, including tube dislocation, blockage, and leakage. Looking at the risk/benefit ratio, we agree with Braga and colleagues that the use of EEN should be targeted selectively at the group of patients with malnourishment, defined as the involuntary weight loss of more than 10% of the patient’s usual body weight in the preceding 6 months. In general, 25 to 30 kcal/kg/day and 1.2 to 1.5 g/kg/day of protein are reasonable goals in these patients (Sampliner, 2001). TPN is used when nutritional goals cannot be met with enteral feedings because of GI dysfunction.

On the other hand, positive effects have been demonstrated by mere institution of early oral intake. In the group randomized to clear liquids ad lib after surgery, followed by a soft diet for patients who can tolerate 500 mL of liquid. Steed and colleagues (2002) showed that length of stay was significantly reduced, but the rates of ileus, emesis, and morbidity remained identical with the control group.

Clear liquids may be served to the patient following extubation on the first postoperative day. Thereafter, the amount and type of oral intake is progressively tailored, according to the patient’s tolerance. Most of our patients are able to tolerate solids by the third or fourth postoperative day. Oral feeding is suspended or reverted back to a liquid-only diet if the patient vomits, and the NG tube is reinserted if the patient vomits on more than one occasion.

Blood glucose level must be closely monitored, with 4 to 6 hourly blood glucose level measurements for the first 24 to 48 hours. For patients who had a TP, blood glucose was regulated with intravenous administration of insulin. After starting oral food intake, patients received either semi–long-acting insulin or a combination of long-acting insulin as baseline therapy with premeal short-acting insulin. This must be done in consult with an endocrinologist and it may entail transfer of the patient to the endocrine unit once surgically fit for discharge. Patients should also be taught self-monitoring of their blood sugar and self-administration of insulin by the diabetic nurse educator.

Following TP, pancreatic enzyme substitutes are also be prescribed to this group of patients to treat their expected exocrine insufficiency. Although many patients complain of some form of diarrhea, this symptom resolves rapidly with high-dose enzyme substitution (Müller et al, 2007; Wagner et al, 2001).

Epidural analgesia or patient-controlled analgesia is continued for the first 72 hours. Oral analgesia is started concurrently once the patient can tolerate oral intake. Active chest physiotherapy with incentive spirometry should commence on the first postoperative day. Abdominal binders are occasionally provided for patients to minimize wound pain during deep-breathing exercises. Patients must be assisted to sit out of their beds on the first postoperative day, and a graded mobilization exercise regimen will allow the patient to begin limb exercises before progressing to supervised mobilization and finally to independent mobilization. Urinary catheters are removed following the discontinuation of the epidural analgesia. All patients who had their spleen removed should be vaccinated against Streptococcus pneumoniae and Hemophilus influenzae type B.