Chapter 61B. Perspective on Surgery for Exocrine Neoplasms of the Pancreas

The last 40 years have seen remarkable advances in what we know about pancreatic neoplasms, their biology, how we approach their management, and the quality and safety of surgical treatment. Some can be attributed to better surgical technique, but most are due to developments in imaging, along with nonsurgical interventions such as endoscopic retrograde cholangiopancreatography (ERCP), stenting, and percutaneous or endoscopic needle biopsies. In most cases the diagnosis can be made preoperatively; the extent of the tumor can be determined; and the timing as well as the nature of the probable surgical procedure can all be planned before or even instead of a laparotomy. Pancreaticoduodenectomy has become sufficiently safe (2–5% mortality, median postoperative length of stay 8 days in high-volume centers) that the operation can be offered to most patients without biopsy proof of malignancy because the risk of missing a cancer now exceeds the risk of mistakenly operating for a benign condition. This radical approach to resection applies as well to ampullary neoplasms, which contain cancer in up to 50% of villous adenomas in spite negative biopsies. The conclusion is that a negative biopsy should not deter resection of a lesion with significant malignant potential.

Yet there are still important shortcomings to the methods we have. There is no reliable screening test for pancreatic cancer and, even if one were invented that had a remarkable 99% accuracy, we would probably be unable to find the lesion at the desired very early stage of growth because current imaging with the best computed tomography (CT) or endoscopic ultrasound (EUS) still cannot “see” masses much smaller than 1 cm (at which size many pancreatic adenocarcinomas have already spread).

For purposes of staging pancreatic cancers, multidetector contrast-enhanced angio-CT is as good as we have for evaluating the mesenteric, celiac, and hepatic blood vessels, but it still misses small liver or peritoneal metastases in at least 10% of cases of apparently resectable cancers.1 In addition, the use of preoperative laparoscopy will in another 10% demonstrate peritoneal dissemination of cancer cells that indicate a significantly diminished prognosis.2

There is a consensus that preoperative biliary stenting is unnecessary if the operation can be expeditiously performed, but it is useful when the patient is uncomfortable with pruritus and relief by biliary decompression will not be immediate or when neoadjuvant treatment is planned. The fear of increased postoperative surgical site infections has not been substantiated in a recent randomized controlled trial.3 Whether neoadjuvant therapy confers a benefit is debatable, either for downstaging borderline resectable cancers involving the mesenteric vessels or for increasing long-term survival.4 The M.D. Anderson group has argued that neoadjuvant chemoradiation helps to define the 25% of patients in whom metastases are predestined to blossom immediately, ensures delivery of the treatment to some patients who would not tolerate or receive postoperative adjuvant treatment, and may improve survival, perhaps by reducing positive resection margins. However, with the relatively ineffective chemotherapy drugs available, the end results of a neoadjuvant approach have not been convincingly superior. In contrast to the prevailing European studies, adjuvant radiation added to chemotherapy for resected pancreatic cancer was associated with a significant survival advantage demonstrated in a large American database.5

Many studies have demonstrated that higher volumes, whether of the surgeon or the hospital, are related not only to lower postoperative mortality but also to higher long-term survival. New data show that both better technique, manifested by a higher percentage of negative margins,6 and better hospital infrastructure for perioperative support play a part. Ensuring an adequate retroperitoneal margin by skeletonizing the superior mesenteric artery of its nerve plexuses (pancreatic cancer spreads along perineural channels) may help to reduce the positive margin rate that otherwise may be as high as 75% when there is assiduous pathological examination.7 Intraoperative radiation, which in theory might be useful for potentially positive margins, has been disappointing in achieving better survival.8

The technique for pancreaticoduodenectomy described in Chap. 60 is acceptably generic. It does not address the growing utilization of laparoscopic techniques for resection, now common for distal pancreatectomy but perhaps in the future for pancreaticoduodenectomy as well.9 It also does not take sides in the debate about whether to perform a “classic Whipple operation” with pyloroantrectomy and gastrojejunostomy or a pylorus-preserving version with a duodenal-jejunostomy. Unless the proximal duodenum or antrum is directly involved by the cancer, there seems to be no demonstrable difference in survival or cure rates, despite that there may be some difference in extent of lymphadenectomy (even extended lymph node dissections have proven to make no difference).10

A recent meta-analysis of pylorus-preserving versus standard pancreaticoduodenectomy with antrectomy has found no significant difference in various perioperative indices except for 275 cc of greater blood loss and 72 minutes longer operation with the antrectomy version.11 They noted no difference in the occurrence of delayed gastric emptying with the pylorus-preserving version. Our experience at the Massachusetts General Hospital differs in that the antrectomy portion adds less than 15 minutes for mobilization and the Hofmeister turn-in of part of the gastric staple line, and we, along with some others, have found that one-third of patients have 5–7 extra days of gastroparesis after the pylorus-preserving operation (erythromycin did not help). In our practice we avoid pylorus preservation. From my perspective, you can take it or leave it.

While extended retroperitoneal lymph node dissection has not fulfilled its theoretical promise, vascular resection, at least portal and mesentery vein resection, is establishing acceptance.12 Involvement of the superior mesenteric or celiac arteries probably signals that the cancer has spread far enough along retroperitoneal lymphatic and neural channels that control of margins and metastases is out of reach, but vein resection when the cancer is adherent or invasive—it is difficult to differentiate between these without resecting the vein—can be accomplished safely with consequent negative margins and a 5-year survival comparable to patients not requiring a vein resection. The extent of vein involvement that still allows resection and reconstruction, whether with a graft or with mobilization of the mesentery and end-to-end anastomosis, will vary with the skill and aggression of the surgeon.

Postoperative leak at the pancreatic anastomosis leading to a pancreatic fistula or intra-abdominal collection is one of the most common complications of a pancreaticoduodenectomy, and the one most likely to be lethal. Breakdown of the pancreaticojejunostomy is more common with a soft, nonfibrotic pancreas, which does not hold sutures as well. It occurs more frequently, therefore, after resections for cystic neoplasms, neuroendocrine tumors, bile duct cancers, and duodenal cancers than in pancreatic cancers, which obstruct the pancreatic duct and cause induration of the gland. Fistulas occurred in 13% of our Whipple operations (75/581) over 5 years.13 While 39% of these fistulas healed with closed-suction drainage, allowing the drains (which we routinely use) to be removed, 61% were high impact and were complicated by an abscess or bleeding requiring intervention. Seven patients died (9% of those with fistulas), 6/7 from vascular erosion and pseudoaneurysm. A sentinel bleed from a drain must be taken very seriously and it warrants immediate angiographic evaluation for purposes of embolization or stenting of the culprit vessel.

Probably all high-volume pancreatic surgery practices are seeing increasing numbers of pancreatic cystic tumors, in large measure the product of cross-sectional imaging for other purposes. Thirty years ago, pseudocysts were said to be the most common cystic lesions of the pancreas; now cystic neoplasms are far more prevalent, the majority being asymptomatic and found by serendipity. Cystic neoplasms comprised about one-fourth of our pancreaticoduodenectomies in 2009 and are now the most common pancreatic neoplasm entering our practice. The challenge for pancreatic surgeons and their colleagues is how to estimate the relative risk of watchful waiting versus intervention.

Much attention is paid to the differential diagnosis of serous cystadenomas (SCA), mucinous cystic neoplasms (MCN), intraductal papillary mucinous neoplasms (main-duct and branch-duct IPMN), cystic neuroendocrine tumors, and other uncommon entities by the use of CT, magnetic resonance imaging (MRI), EUS, and fine-needle aspiration, the latter in part for cytology but especially for carcinogenic embryonic antigen (CEA) levels in the fluid. An elevated CEA tends to indicate one of the mucinous family of tumors but gives no indication of malignancy at any level.14 However, the level of sensitivity and specificity of these diagnostic tests generally does not exceed 70% across the board, with the exception that an elevated CEA essentially excludes serous cystic neoplasms, which are almost always benign.

From a pragmatic standpoint, therefore, it may be more useful to develop guidelines that can be applied to an undetermined cystic lesion that does not fit the clinical or morphologic criteria for a pseudocyst. Some are easy: if the cyst is producing symptoms (pain, jaundice, and pancreatitis) or is bulky and demonstrably growing over time, it probably should be resected. If the main pancreatic duct is dilated, main-duct IPMN, with its 60% likelihood of containing at least in situ cancer, must be considered and resected. If there is a mural nodule or solid component, the risk of malignancy is too great to ignore. The Sendai consensus,15 which provides reliable guidelines that have been repeatedly validated, adds the element of cyst size. While there are uncommon exceptions, generally there is minimal risk of malignancy, even carcinoma-in-situ, in mucinous cysts (MCN or branch-duct IPMN) smaller than 3 cm. Consequently smaller, asymptomatic cysts without mural nodules or other solid components can be watched for growth or change, preferably by MRI/magnetic resonance cholangiopancreatography (MRCP), perhaps every year or two. Most will turn out to be branch-duct IPMNs, at least 30% of which will be multiple. Those individual cysts that fit a guideline for potential malignancy need segmental resection—metachronous small cysts can be left behind and watched for future growth and change.

There is growing appreciation that many small, incidentally discovered pancreatic cystic lesions do not grow or develop cancerous change over a period of many years while many, perhaps 5% (8/159 in our experience16), turn out to be indeterminate nonneoplastic cysts upon resection. On the other hand, because the mucinous cystic neoplasms—MCN or IPMN—may eventually degenerate into aggressive incurable adenocarcinomas, there is a premium on timely determination of the need for resection, erring on the side of taking out many benign lesions to prevent one cancer from getting beyond cure. Cystic carcinomas are highly curable if removed before invasion or metastases, and the majority of those that are invasive can still be cured. Lymph node metastases are less common in IPMN carcinomas and unheard of in MCN carcinomas, at least in our experience.17 For this reason segmental and other atypical resections that focus on the locale of the lesion suffice. Local excision of the tumor, including duodenum-preserving head resection (which is essentially a wide enucleation), is also acceptable as long as the tumor can be removed with adequate surrounding pancreas to ensure negative margins. Evaluation of the pancreatic duct margin, of special importance for main-duct IPMN, requires an experienced pathologist to differentiate mucinous hyperplasia (acceptable) from neoplasia (unacceptable) on frozen section.

Surveillance, generally with MRI/MRCP, is indicated for residual cysts, perhaps no more often than at intervals of 1 or 2 years. We have not seen the appearance of a new serous or mucinous cystic neoplasm after resection of the propositus lesion and do not recommend surveillance imaging for these patients. Surveillance is, of course, appropriate after removing a cancer or any main-duct IPMN because of the possibility of metastasis, recurrence, or development of a new main-duct lesion in the residual pancreas. In addition, patients with IPMN are at somewhat greater risk of developing other gastrointestinal cancers, especially in the colon and stomach.