INTRODUCTION  Cancer of the exocrine pancreas is the fourth leading cause of cancer-related death in the United States, and is second only to colorectal cancer as a cause of digestive cancer-related death. Surgical resection is the only potentially curative treatment. Unfortunately, because of the late presentation of the disease, only 15 to 20 percent of patients are candidates for pancreatectomy. The prognosis of pancreatic cancer is poor even in those with potentially resectable disease. The five-year survival following pancreaticoduodenectomy is only about 25 to 30 percent for node-negative, and 10 percent for node-positive tumors.

More recent data suggest that outcomes may be improving over time, possibly related to an increased proportion of patients undergoing surgery at teaching hospitals, and/or greater use of adjuvant therapy. As an example, in a report of 396 Medicare patients residing in one of 11 SEER (Surveillance, Epidemiology, and End Results reporting) registries who underwent potentially curative resection of pancreatic cancer between 1991 and 1996, the three-year survival rate was 34 percent  . In multivariate analysis, one of the strongest predictors for survival was the use of adjuvant chemoradiotherapy; the three year survival rate was significantly higher among those who received it compared to those who did not (45 versus 30 percent, respectively).

The epidemiology, clinical features, diagnosis, and surgical staging of pancreatic exocrine cancer will be reviewed here. These topics are also part of an official guideline issued by the American Gastroenterological Association.

EPIDEMIOLOGY  In the United States, the incidence of pancreatic cancer has been on the rise since the 1930s, but it has been relatively stable since the 1970s. Approximately 33,730 new cases are anticipated in 2006, with 32,300 expected deaths . Mortality rates closely follow incidence rates because of the poor prognosis. Pancreatic cancer is rare before the age of 45, but incidence rises sharply thereafter.

A database study in the United States estimated that the age-adjusted incidence rates were approximately 30 percent higher in men and 50 percent higher in blacks compared with whites and people of other races . During a period between 1977 and 2001, the proportion of patients with localized disease decreased from 12 to 7 percent and the proportion of those with regional disease increased from 19 to 26 percent while the proportion of patients with metastatic disease remained constant (about 50 percent).

Risk factors  The risk factors for and molecular pathogenesis of pancreatic cancer are discussed in detail elsewhere. Summarized briefly, the major risk factors include chronic pancreatitis, smoking, diabetes mellitus, and hereditary predisposition to pancreatic cancer itself or to multiple cancers.

CLINICAL FEATURES  A diagnosis of pancreatic cancer is often suspected on the basis of the history; there are few confirmatory physical findings, with the exception of jaundice.

History  Most patients with pancreatic cancer experience pain, weight loss, or jaundice .

• Pain is present in 80 to 85 percent of patients with locally advanced or advanced disease . The pain is usually felt in the upper abdomen as a dull ache that radiates straight through to the back. It may be intermittent and made worse by eating.

• Weight loss can be profound; it may be associated with anorexia, early satiety, diarrhea, or steatorrhea.

• Jaundice is often accompanied by pruritus, acholic stools, and dark urine. Painful jaundice is present in approximately one-half of patients with locally unresectable disease, while painless jaundice is present in approximately one-half of patients with a potentially resectable and curable lesion .

The initial presentation varies according to tumor location. Tumors in the pancreatic body or tail usually present with pain and weight loss, while those in the head of the gland typically present with steatorrhea, weight loss, and jaundice. The recent onset of atypical diabetes mellitus, a history of recent but unexplained thrombophlebitis , or a previous attack of pancreatitis are sometimes noted.

A population-based case control study included 120 patients with pancreatic cancer whose symptoms were compared with 188 age- and sex-matched controls . Most signs and symptoms occurred within three years prior to diagnosis and were non-specific. On multivariable analysis, symptoms and signs more likely to be reported by cases included abdominal pain, appetite loss, jaundice, pale-colored stools, unusual belching, weight loss, and unusual bloating.

Physical findings  An abdominal mass or ascites is noted at presentation in 20 percent of patients with pancreatic cancer. A nontender but palpable gallbladder may be seen or felt at the right costal margin in those with jaundice. Left supraclavicular lymphadenopathy (Virchow's node) or a palpable rectal shelf are present in some patients with widespread disease. Rarely, subcutaneous areas of nodular fat necrosis (pancreatic panniculitis) may be evident.

DIAGNOSIS AND STAGING  A variety of diagnostic studies are available for the diagnosis and staging of pancreatic cancer, with varying sensitivity and specificity. Routine laboratory tests may reveal a rise in the serum bilirubin concentration and alkaline phosphatase activity, and the presence of mild anemia. The diagnosis of pancreatic cancer, however, is typically made radiographically and histologically. The differential diagnosis includes chronic pancreatitis, pancreatic endocrine tumors, autoimmune pancreatitis, lymphoma, and a variety of other rare conditions.

Imaging studies The diagnosis of pancreatic cancer is usually based upon the results of one or more of the following imaging studies: transcutaneous or endoscopic ultrasound (US or EUS); computed tomography (CT); endoscopic retrograde cholangiopancreatography (ERCP); magnetic resonance imaging (MRI), and MR cholangiopancreatography (MRCP). Many of these studies are also useful for staging in order to determine tumor operability.

  Ultrasound The initial study in patients who present with jaundice is usually abdominal US . Dilated bile ducts or the presence of a mass in the head of the pancreas suggest the presence of a pancreatic tumor. The reported sensitivity and specificity of US in diagnosing pancreatic cancer is 75 to 89 and 90 to 99 percent, respectively; however, these numbers are dependent upon the expertise of the ultrasonographer, the presence or absence of bile duct obstruction, and the extent of the tumor .

  CT and CT angiography  CT scan has a better sensitivity than, and similar specificity to US (85 to 90 and 90 to 95 percent, respectively) for the detection of pancreatic cancer . It may be particularly useful in patients who are not jaundiced, and in those in whom intestinal gas interferes with US. CT may reveal bile and pancreatic duct dilation, a mass lesion within the pancreas, and evidence of extrapancreatic spread including liver or lymph node metastases and ascites.

Helical CT combined with a bolus administration of intravenous contrast (CT angiography) can provide useful information regarding major vessel involvement (eg, portal and superior mesenteric vein and superior mesenteric artery) that reflects surgical unresectability . One series evaluated 48 major vessels in 25 patients with pancreatic cancer using both spiral CT and operative dissection to determine resectability . Tumor contiguity to the portal vein, superior mesenteric artery or vein, hepatic artery, or celiac axis (with invasion of <25 percent of the vessel circumference) was associated with resectability in all cases. In contrast, the presence of periportal collaterals, dilated small peripancreatic veins, or an otherwise unexplained portal vein thrombosis suggests portal vein occlusion and a high likelihood of unresectability .

Helical CT may underestimate the degree of hepatic or lymph node involvement; metastatic cancer may be found in normal size lymph nodes, and CT frequently misses hepatic lesions <1 cm in size. Furthermore, enlarged peripancreatic lymph nodes can be benign, and their presence does not correlate with either metastatic cancer or survival .

  ERCP  ERCP is widely used as a diagnostic tool for pancreatic cancer; it has a sensitivity and specificity of 90 to 95 percent. It is most useful if the CT or US does not reveal a mass lesion or if chronic pancreatitis is in the differential diagnosis. Findings suggestive of a malignant tumor include superimposable strictures or obstruction of the common bile and pancreatic ducts (the "double duct" sign), a pancreatic duct stricture in excess of 1 cm in length, pancreatic duct obstruction, and the absence of changes suggestive of chronic pancreatitis.

Some of the limitations of ERCP are that parenchymal abnormalities can only be detected by inference, that tumors can be missed in the uncinate process, accessory duct, and tail, and the need for intraductal contrast administration. Therapeutically, ERCP is indicated for patients who present with cholangitis, or for those who require relief of biliary obstruction for palliation.

  EUS  A number of reports have evaluated the accuracy of EUS in diagnosing and staging pancreatic cancer. The results have been variable, in part reflecting differences in the gold standard against which EUS was compared . EUS is very operator-dependent; as a result, its value varies with locally available expertise. As general rules, EUS appears to be most useful for diagnosis of small tumors (eg, less than 2 to 3 cm in diameter) and may also be helpful for evaluating the possibility of nodal and major vascular involvement except for the superior mesenteric vein and artery. This topic is addressed in detail elsewhere.

  MRI and MRCP  Routine MRI offers no significant diagnostic advantage over CT for the staging workup of pancreatic cancer . Combining the two offers little that cannot be achieved with one alone. Thus, the choice of MRI or CT depends upon the level of locally available expertise and the clinician's comfort with one or the other radio-imaging technique. We prefer helical CT angiography.

On the other hand, MRCP is a useful adjunct to other radiographic diagnostic techniques, and may emerge as the preoperative imaging procedure of choice for these patients. MRCP uses MR technology to create a three dimensional image of the pancreaticobiliary tree, liver parenchyma, and vascular structures. MRCP is better than CT for defining the anatomy of the biliary tree and pancreatic duct, has the capability to evaluate the bile ducts both above and below a stricture, and can also identify intrahepatic mass lesions. It is at least as sensitive as ERCP in detecting pancreatic cancers , and unlike conventional ERCP, does not require contrast material to be administered into the ductal system. Thus, the morbidity associated with endoscopic procedures and contrast administration is avoided.

Although MRCP has not yet replaced ERCP in the patients suspected of having pancreatic cancer in all centers, it may be preferred in specific settings:

• Patients who have gastric outlet or duodenal stenosis or who have had surgical rearrangement (eg, Billroth II) or ductal disruption, resulting in ducts which cannot be assessed by ERCP.

• To detect bile duct obstruction occurring in the setting of chronic pancreatitis.

• For patients in whom attempted ERCP is either totally unsuccessful or provides incomplete information because of pancreatic duct obstruction .

Serum tumor markers Several serum markers for pancreatic cancer have been evaluated, the most useful of which is cancer associated antigen 19-9 (CA 19-9). It has a reported sensitivity and specificity of 80 to 90 percent, respectively . However, these values are closely related to tumor size. The accuracy of CA 19-9 to identify patients with small surgically resectable cancers is limited

Furthermore, CA 19-9 is frequently elevated in patients with various benign pancreaticobiliary disorders . One study found that serum concentrations above 37 U/mL represented the most accurate cutoff value for discriminating pancreatic cancer from benign pancreatic disease (sensitivity and specificity of 77 and 87 percent, respectively) . As a result of all of these issues, CA 19-9 is not recommended as a screening test for pancreatic cancer.

Very high CA 19-9 levels (eg, >1000 U/mL) are often associated with surgically unresectable lesions, and higher baseline serum levels of CA 19-9 are linked to inferior prognosis. In one report of 154 patients receiving chemotherapy for inoperable disease, one year survival was significantly lower among those with a pretherapy CA 19-9 level >958 U/mL (19 versus 46 percent) . Nevertheless, an expert panel convened by the American Society of Clinical Oncology (ASCO) recommended against the use of CA 19-9 alone as an indicator of operability .

Serial monitoring of CA 19-9 levels (once every one to three months ) is useful to follow patients after potentially curative surgery and for those who are receiving chemotherapy for advanced disease. Rising CA 19-9 levels usually precede the radiographic appearance of recurrent disease, but confirmation of disease progression should be pursued with imaging studies and/or biopsy

Although several marker candidates have emerged from preclinical studies (and one, macrophage inhibitory cytokine-1, appears promising, none has replaced CA 19-9 to date.

Cytology  Percutaneous fine needle aspiration (FNA) biopsy of a pancreatic mass can be performed using either US or CT guidance. The sensitivity and specificity of this procedure for the diagnosis of pancreatic cancer depends upon tumor size and operator expertise; values in the range of 80 to 90 and 98 to 100 percent, respectively, have been reported .

A theoretical concern is that percutaneous FNA biopsy of the pancreas may disseminate tumor cells intraperitoneally or along the needle path in patients who are believed to be candidates for potentially curative resection. However, the risk appears to be quite low or absent. In one study of 41 patients undergoing resection for primary pancreatic adenocarcinoma, 21 of 32 patients without preoperative open biopsies had undergone preoperative CT or fluoroscopically guided FNA . There was no increase in positive peritoneal washings, peritoneal failure rate, or median survival in these patients. Nevertheless, concern persists and, in practice, we try to avoid percutaneous FNA in patients with resectable masses.

EUS-guided FNA biopsy is increasingly being used as a tool for both the diagnosis and staging of pancreatic cancer, even in patients who otherwise have resectable tumors . This procedure is less likely to cause intraperitoneal spread of the tumor since the biopsy is obtained through the bowel wall rather than percutaneously . In one study, 44 patients had EUS-guided FNA of pancreatic lesions (39 head/neck, 5 body, 3 tail) and associated lymph nodes . Adequate specimens were obtained in 44 of 47 pancreatic lesions (94 percent) and 14 of 14 associated lymph nodes. Of the 46 lesions in which specimens were adequate and a final diagnosis was available (32 malignant, 14 benign), EUS-guided FNA had a sensitivity of 92 percent, specificity of 100 percent, and diagnostic accuracy of 95 percent for pancreatic lesions; the comparable values for lymph node involvement were 83, 100, and 88 percent, respectively. Overall, EUS findings in combination with EUS-guided FNA precluded surgery in 12 of 44 patients (27 percent) and avoided the need for further diagnostic tests, thereby expediting therapy in 25 patients (57 percent). Clinical decisions were influenced in two-thirds of cases.

Samples of pancreatic juice obtained from the duodenum or pancreatic duct and those obtained by brushing ductal lesions at the time of ERCP can also be sent for cytology. The sensitivity of these techniques is poor, but the specificity approaches 100 percent.

The addition of molecular genetic analysis (eg, assay for K-ras gene mutations by polymerase chain reaction or restriction fragment length polymorphism) to cytologic examination may improve sensitivity, especially in patients with small primary tumors. In one study for example, evidence of K-ras gene mutations was found in the pancreatic juice in 11 of 12 adenocarcinomas (92 percent), versus no samples of pancreatitis or islet cell tumors . Four of the original 20 samples (20 percent) could not be analyzed due to the absence of cells in the aspirate.

Staging laparoscopy  Laparoscopy alone or with laparoscopic ultrasonography is emerging as a new staging modality for periampullary tumors (ie, those involving the head of the pancreas, lower one-third of the bile duct, ampulla of Vater, and first three parts of the duodenum). Small occult metastases (<1 cm in diameter) on the surface of the liver or peritoneum which are rarely visible by CT, MRI, or transabdominal US may be visualized by this modality. If the tumor appears to be resectable following laparoscopic visualization, laparoscopic ultrasonography may be performed; the liver and portal, peripancreatic, periaortic, and celiac axis lymph nodes are evaluated, and possible vascular involvement can be determined.

The major benefit of staging laparoscopy is in reducing the number of apparently operable cases that are found to be unresectable at laparotomy . The efficacy of this approach was illustrated in a series that evaluated staging laparoscopy and laparoscopic ultrasound in 40 patients with periampullary tumors. Previously unsuspected metastases were discovered by laparoscopy in 14 patients (35 percent) in the liver, on peritoneal surfaces, or in portal lymph nodes. Laparoscopic ultrasonography identified other indications of unresectability (eg, vascular invasion or liver metastases) in an additional 10 patients (25 percent). Using this combined approach, 11 of 14 tumors considered to be resectable after staging were actually resectable at the time of laparotomy, while 23 of 24 tumors deemed unresectable were confirmed by histology or subsequent exploration. Overall, the sensitivity, specificity, and accuracy of staging laparoscopy and ultrasonography were 92, 88, and 89 percent, respectively.

I use staging laparoscopy selectively in patients with advanced vascular involvement (but not yet complete encasement or occlusion of the major vessels) and those with pancreatic body or tail lesions who are not jaundiced

Surveillance in high-risk patients  Up to 10 percent of patients with pancreatic cancer have a family history of the disease The best surveillance strategy for such patients are uncertain. In one report, a program of ERCP, spiral CT, and determination of serum carcinoembryonic antigen and CA 19-9 concentrations detected pancreatic dysplasia without cancer in 7 of 14 patients who had two or more family members with pancreatic cancer in at least two generations . Screening with EUS can detect resectable lesions, but has a high false-positive rate .

The optimal timing and frequency of screening for pancreatic cancer in individuals at risk is uncertain. The American Gastroenterological Association (AGA) recommends that screening begin at age 35 for those with hereditary pancreatitis, and 10 years before the age at which pancreatic cancer was first diagnosed in individuals with a positive family history . Unfortunately, currently available techniques, even invasive tests, lack sensitivity for the detection of very small lesions. The AGA recommends the use of spiral CT and EUS.

RECOMMENDATIONS Patients with pancreatic cancer usually come to attention because of jaundice, unexplained upper abdominal pain/weight loss, or an unexplained episode of pancreatitis. The initial study performed is typically either a conventional US or CT scan and the initial questions that must be answered are whether pancreas is the cause of the symptoms and whether the lesion is benign or malignant

Mass lesion in a patient who is unfit for a major pancreatic resection  For patients who are unfit to undergo a major resection, a tissue diagnosis should be made to permit palliation by endoscopic, chemotherapeutic, or radiotherapeutic means. This can be accomplished either by percutaneous (CT or US guided) fine needle aspiration biopsy or by endoscopically obtained brushings or biopsies. Endoscopic biopsy can frequently be aided by EUS guidance.

Mass lesion not seen on CT or ultrasound  When no mass lesion is detected on CT or ultrasound, an ERCP should be performed to exclude biliary tract stones or some other non-neoplastic cause of symptoms. The ERCP may also confirm the suspected diagnosis of pancreatic tumor if a periampullary mass is detected or if malignant appearing strictures of the pancreatic and/or bile duct are found.

Mass lesion in a candidate for major pancreatic resection  When a mass lesion of the pancreas is detected on CT or US, it is reasonable to conclude that a neoplasm (most likely malignant) is present and no further diagnostic tests are needed for patients who are otherwise fit to undergo a major pancreatic resection. Attempts to make a tissue diagnosis in such patients are not useful since a benign sample does not exclude the presence of a neighboring malignancy and, in the absence of a diagnostic sample, one must assume that the lesion is malignant. Attempts to make a preoperative tissue diagnosis may in fact be detrimental if tumor cells are disseminated during percutaneous biopsy. Thus, for surgically fit patients, we operate without attempting to establish a preoperative diagnosis of malignancy. However, it must be recognized by both the clinician and the patient that uncertainty regarding the diagnosis in these instances persists and that some patients with benign lesions may be subjected to the radical resections that might be more appropriate for malignant lesions.

Focal chronic pancreatitis and autoimmune pancreatitis are the two benign processes most commonly mistaken for pancreatic malignancy on the basis of CT or US. In fact, the majority of patients found to have benign disease after undergoing radical resection for presumed malignancy suffer from one of these conditions. These diagnoses can sometimes be suspected on the basis of history (eg, extreme young age, prolonged ethanol abuse, history of other autoimmune diseases). In this subset of patients, further imaging studies (either ERCP or MRCP), may be particularly helpful if they reveal multifocal biliary strictures (suggestive of autoimmune pancreatitis) or diffuse pancreatic ductal changes (suggestive of chronic pancreatitis). The frequency of radical resection for benign disease may be reduced when this approach is combined with EUS-guided transduodenal biopsy.

  Determining resectability  The next issue for patients who have a mass lesion and who are surgical candidates is determining resectability. Most surgeons limit radical resections to the management of patients with potentially curable lesions. This usually requires that the tumor does not involve sites that would not be encompassed within the resection and that the tumor does not involve adjacent critical vascular structures such as the SMA/SMV, portal vein, celiac axis, or hepatic artery.

There are two approaches that might be taken to resolve this issue: a traditional approach and an alternative approach.

  Traditional approach  The traditional approach involves a contrast-enhanced helical CT scan with timed image sequences that permit evaluation of vascular structures and subtle metastatic implants (ie, CT angiography with phased images and thin cuts). In our own series of 100 patients with periampullary neoplasms evaluated by helical CT and CT angiography, the lack of major vascular involvement by tumor was associated with a resectability rate of over 90 percent. Partial involvement of the SMV and/or SMA on CT angiography was associated with a resectability rate of 10 to 50 percent depending upon the extent of the vascular involvement.

Based upon this experience, our approach depends strongly upon the findings of CT angiography. For jaundiced patients with no involvement or minimal involvement of the major vessels and no evidence of distant metastases, we proceed directly to an attempt at surgical resection. For non-jaundiced patients (ie, those with body or tail tumors), or those with major but incomplete involvement of the vascular structures, we perform preoperative laparoscopy to exclude tiny metastases that might have been overlooked by CT. If the laparoscopy was negative, we then embark on a radical surgical resection. Finally, for those with circumferential involvement of one or more of the major vessels, we consider the lesion unresectable and proceed to obtain tissue for diagnosis and palliative non-operative treatment.

  Alternative approach  Management of patients using the above described "traditional approach" will result in operating on some 10 to 15 percent of patients with tumors that cannot be considered surgically "curable" (ie, they are found to invade the critical peripancreatic vascular structures, distant organs, or distant nodes at the time of attempted resection). Some have advocated performance of additional preoperative "staging" studies to identify such patients and spare them an unnecessary laparotomy. According to this alternative approach, patients with lesions felt to be potentially curable on the basis of helical CT angiography are studied further by MRCP, EUS, laparoscopy, and/or laparoscopic US.

The role of EUS in the management and preoperative staging of these patients is more controversial. A number of studies have indicated that EUS may be as good or even better than CT angiography in staging pancreatic cancer and in determining if pancreatic tumors are resectable for cure. However, in our experience, EUS has not proven to be better at establishing resectability when compared to the high quality CT angiography. Thus, the choice between CT angiography and EUS probably depends upon the locally available expertise.

The ultimate role of laparoscopy and laparoscopic ultrasonography in the management and staging of these patients has not been defined. Experience with laparoscopic ultrasonography in the management of pancreatic cancer is limited (see above). As a result, its role remains to be determined. Our experience with laparoscopy as a method of identifying surgically incurable patients prior to laparotomy has been disappointing in jaundiced patients with pancreatic head lesions. Attempts at resection are usually aborted because of local or vascular extension of the disease rather than because of laparoscopically detectable metastatic lesions. Laparoscopy performed in approximately 100 patients at our institution demonstrated unresectability because of metastatic disease in only 10 percent of those eventually found to be unresectable at laparotomy. The remaining patients were found to be unresectable because of either major vascular involvement, hilar adenopathy, or intraparenchymal metastases that would have been overlooked during laparoscopy.

  Conclusion  Although both the so-called "traditional" and "alternative" approaches are supportable by the available evidence, our practice is to follow the traditional approach in determining the surgical resectability of a suspected pancreatic cancer.