Epidemiology, Causal Factors, Genetics, and Cytogenetic Abnormalities

Pancreatic cancer is the fifth leading cause of cancer mortality among men and women of all ages; it strikes an estimated 37,680 Americans annually and caused 34,290 deaths in 2008.¹ A National Cancer Institute (NCI) survey covering 1973 to 1988 showed less than 20% of affected patients survived 1 year and only 3% were alive 5 years after diagnosis.^2,3 Even with the addition of adjuvant radiation and chemotherapy as well as recent decreases in surgical morbidity, only very modest improvements in survival have been realized. Worldwide, the reported incidence varies considerably. It is unclear whether this represents a true phenomenon or whether it is a reflection of more meticulous reporting patterns in more industrialized countries. In the United States, the incidence of pancreatic cancer is 9 in 100,000 in whites and 15.2 in 100,000 in blacks with a male-to-female ratio of 1.3 : 1. Diagnosis is rare before age 45 but rises sharply thereafter.² Although the cause of pancreatic cancer is unknown, several environmental factors have been implicated. Smoking, in particular, raises the relative risk 1.5 times.⁴ Diets high in meat or fat also have been linked to increased risk, and a diet of fresh fruits and vegetables has been found to be protective.⁵ An increased incidence also is found with a prior history of surgery for peptic ulcer disease.⁶

Much controversy was generated in the early 1980s with a report linking pancreatic cancer with coffee consumption,⁷ but subsequent prospective studies have failed to confirm this.⁸ Both diabetes mellitus^7,9 and chronic pancreatitis¹⁰ are associated with pancreatic cancer, but it is unclear whether this is a cause-and-effect relationship. Workers employed in manufacturing 2-naphthylamine, benzidine,¹¹ and gasoline¹² are reported to have a fivefold increased risk. More recently, the study of deoxyribonucleic acid content, oncogenes, and molecular biology has yielded intriguing data.¹³ Molecular histology data have been nicely summarized by Hruban.¹⁴ More and more evidence paints a picture of pancreatic cancer as a disease of acquired and inherited mutations. It was not until very recently that an international system of nomenclature was adopted that reflected the development of pancreatic ductal carcinoma as a continuum analogous to changes seen in the development of cervical carcinoma.¹⁵ Pancreatic intrepithelial neoplasia 1-3 is used to grade noninvasive early changes of pancreatic ductal neoplasia. Tumor suppressor genes regulate cell proliferation and their inactivation can play a profound role in the development of malignancies. There are four specific genes identified as crucial in the development of pancreatic cancer: p16, p53, DPC4, and BRCA2. An increased incidence of pancreatic cancer has been seen in families with the breast cancer-associated BRCA2 gene.¹⁶ Oncogenes become active by mutation or amputation. An example is the k-ras oncogene, which is often activated in human pancreatic duct carcinomas.^17–21 Data from Almoguera and colleagues²¹ demonstrated that 95% of pancreatic adenocarcinomas contain k-ras oncogenes activated by a mutation at codon 12. This mutation is critical to oncogenesis and may mediate resistance to epidermal growth factor receptor inhibitors. Shibata and associates¹⁸ noted similar findings in 72% of patients undergoing fine needle aspiration. Kondo and colleagues²² also found specific mutations of the k-ras oncogene at codon 12 in 67% of pure pancreatic juice collected endoscopically in patients with proven pancreatic carcinoma. K-ras abnormalities were not detected in healthy controls, in patients with chronic pancreatitis, or in three patients with islet cell tumors.¹³

A summary of genetic abnormalities is provided in Table 38-1. The identification of consistent genetic abnormalities and the observation of higher cancer rates in some families were the genesis of the Johns Hopkins University Hospitals’ National Familial Pancreas Tumor Registry. Familial cases are defined as those in which two or more first-degree relatives are affected. In these individuals the risk extends to second-degree relatives who have a 3.7% chance of developing pancreatic cancer as opposed to 0.6% in second-degree relatives of sporadic cases.²³

Table 38-1 Genetic Abnormalities in Pancreatic Cancer

Anatomy

The pancreas is an elongated, coarsely lobulated gland lying transversely and retroperitoneally in the posterior abdomen at approximately the L1 to L2 level (Fig. 38-1A). The head lies in the duodenal flexure on the right and the tail extends to the spleen. The pancreas is divided into head, uncinate process (considered to be part of the head), neck, body, and tail. Tumors arising to the right of the superior mesenteric vein (SMV) are considered to be in the head, those arising to the left of that vein and left border of the aorta are part of the body, and those arising from the tail are located between the left border of the aorta and the hilum of the spleen. The pancreas is in close contact with surrounding organs including the spleen, stomach, duodenum, jejunum, and kidneys, all of which may be involved with tumor early on (see Fig. 38-1B). A rich lymphatic network surrounds the pancreas, including the celiac, superior and inferior pancreaticoduodenal, superior mesenteric, porta hepatis, and pancreaticosplenic nodes (Fig. 38-2A). With posterior extension of tumor, the lateral aortic nodes are also at risk. The main venous drainage of the pancreas is via the portal system to the liver. Tumors of the head of the pancreas often cause jaundice caused by invasion or compression of the common bile duct.

FIGURE 38-1 • A, The five parts of the pancreas in relation to the duodenum and major vessels. The line dividing the body and tail is arbitrary. B, Oblique transverse cross-section of the upper abdomen along the long axis of the pancreas corresponding approximately to the levels indicated on the inset diagram. In this view relationships of the pancreas to surrounding dose-limiting structures are quite evident. a, Artery; IVC, inferior vena cava; pt, part; v, vein.

FIGURE 38-2 • A, Lymphatic drainage of the pancreas. Labels indicate representative lymph nodes in the major regional nodal groups. B, Lymph node group involvement in 21 patients with duct cancer in the head of the pancreas. APD, Anterior pancreaticoduodenal node; CBD, common bile duct; Col, midcolic node; GC, greater curvature; IB, inferior body; IH, inferior head; Je, jejunal node; LC, lesser curvature; PPD, posterior pancreaticoduodenal node; Py, pyloric node; S, splenic node; SB, superior body; SH, superior head.

(From Cubilla A, Fortner J, Fitzgerald P: Lymph node involvement in carcinoma of the head of the pancreas, Cancer 41:880, 1978.

Pathologic Conditions

The most common type of pancreatic cancer is of ductal origin, comprising from 75% to 90% of patients.²⁴ It is twice as common in the head as in the body or tail. Perineural invasion occurs in approximately 90% of patients, and in more than 85% of patients tumor has extended beyond the organ at the time of diagnosis. The most common extralymphatic sites of involvement are the liver and peritoneum; the lung is the most commonly involved extra-abdominal organ.

Less frequently occurring exocrine tumors, such as cystadenocarcinoma or intraductal carcinoma, are more common in women and may run a much more indolent course,^25,26 remaining localized for many years and having up to a 50% five-year survival rate.²⁷ Solid and cystic papillary neoplasms, also known as Hamoudi tumors, occur in women in their third decade of life, only rarely metastasize, and have a good prognosis. Rare acinar cell cancers are associated with fat necrosis and high lipase production and have a poor prognosis; they may be associated with a clinical picture that includes rash, eosinophilia, and polyarthralgia. Giant cell tumors, which account for only a small percentage of pancreatic cancers, are very large and aggressive and have a very poor survival rate.²⁸ Clinical presentation of these less aggressive histologies, however, is indistinguishable from the more aggressive pathologic subtypes.²⁵ Cancer metastatic to the pancreas is a relatively uncommon finding. In a series of 2587 consecutive autopsies at the Memorial Sloan-Kettering Cancer Center (MSKCC), 10% of patients were found to have metastases in the pancreas with the most frequent primary sites being breast, lung, or melanoma.²⁴ The remaining 5% of pancreatic cancers are of endocrine origin and are discussed separately.

Clinical Presentation

More than 80% of patients present with pain, jaundice, or both^29,30; these symptoms, along with weight loss, constitute a classic triad. Tumors in the body and tail may not cause jaundice or pain until far advanced; thus their prognosis is particularly poor,³¹ whereas periampullary cancers have a significantly better prognosis³² because of earlier clinical presentation. Tumors of the endocrine pancreas may present in a similar fashion but are often recognized because of various clinical syndromes produced by circulating polypeptides elaborated by the tumor.³³ Rarely, presenting symptoms may be diabetes mellitus⁹ or acute pancreatitis.¹⁰ Infrequently, patients may present with migratory thrombophlebitis (Trousseau sign) or with a palpable gallbladder (Courvoisier sign). Abnormal laboratory tests, such as alteration in glucose, amylase, lipase, bilirubin, alkaline phosphatase, lactate dehydrogenase, or aspartate aminotransferase, may all be present but also are all nonspecific. The search for screening tests to allow for earlier diagnosis and, potentially, to effect cure has generally been elusive with the exception of the Cancer of the Pancreas 2 Screening Project. High-risk but asymptomatic individuals were assessed using endoscopic ultrasound (EUS), and the detection of premalignant and preinvasive lesions suggests that EUS may play a role in screening.³⁴

Routes of Spread

Most pancreatic cancers are unresectable at presentation and thus there is sparse literature systematically analyzing patterns of failure. Cubilla and associates³⁵ reported on the incidence of nodal involvement in completely resected early cancers of the head of the pancreas (see Fig. 38-2B) and showed that carcinomas in the head tended to spread to multiple nodal groups. In addition to initial spread to regional lymph nodes, early peritoneal seeding may be exceedingly common. A series from Massachusetts General Hospital (MGH) reports a 40% incidence of peritoneal studding in patients clinically considered to be resectable and who first underwent laparoscopy before definitive resection.³⁶ In a patient who has undergone curative therapy, once the therapy fails, local recurrence again dominates the clinical picture.^37,38 In a series of Griffen and associates,³⁸ only 17% of patients developed disease purely outside the abdominal cavity (Fig. 38-3).

FIGURE 38-3 • Pancreatic adenocarcinoma and dense fibrosis surrounding central pancreatic duct. Hematoxylin and eosin stain; ×16.

Diagnostic and Staging Studies

Preoperative staging of pancreatic cancer to detect resectability is extremely important to avoid unnecessary surgery. Table 38-2 reflects the staging system as revised in 2010, with the T classification distinguishing between potentially resectable (T₃) and unresectable (T₄) primary tumors. The stage grouping allows stage III to signify unresectable, locally advanced pancreatic cancer; stage IV is reserved for patients with metastatic disease. Pancreatic cancer is considered resectable if there is no portal vein or celiac axis involvement and no metastasis evident. Borderline resectability may include unilateral or bilateral SMV/portal vein impingement, less than 180° tumor abuttment, or encasement of the hepatic artery or SMV if the surgeon determines it is reconstructible.³⁹ Magnetic resonance imaging (MRI), ultrasound, and computed tomography (CT) (Fig. 38-4) can detect pancreatic cancers as small as 1 cm. All three modalities easily detect dilatation of the pancreatic and bile ducts as well as liver metastases.⁴⁰ Ultrasound can help distinguish between obstructive and nonobstructive jaundice, but CT is more helpful in providing definition of the tumor and surrounding structures. A clinical suspicion of pancreatic cancer or evidence of a dilated pancreatic duct warrants a pancreatic CT scan obtained via a defined pancreas protocol such as a helical or spiral triphasic crosectional CT with thin slices. The quality of modern imaging is so good as to predict resectabilty in 80% to 85% of cases when appropriate criteria are met.^41,42 If cancer is fairly certain, the scan should include the chest to rule out metastatic lung disease. Positron emission tomography (PET) is considered a complementary study, as is EUS, which is of particular usefulness in the assessment of vascular invasion and thus resectability.^43,44 Specifically, PET scan is helpful in characterizing equivocal lesions on CT and MRI, but is not helpful as a staging tool.⁴⁵ Advances in technology have improved both the sensitivity and predictive value of cross-sectional imaging to such an extent that abdominal ultrasound, cholangiography, and angiography are no longer required for accurate staging. Percutaneous ultrasound or CT-guided biopsy can safely and easily provide pathologic diagnosis in approximately 90% of cases,⁴⁶ but in the case of adenocarcinoma it may cause increased risk of local failure secondary to rapid seeding of tumor.⁴⁷ A review from MGH⁴⁸ showed a fourfold increased risk of positive peritoneal cytology in patients who had prior percutaneous fine needle aspirates. Some institutions thus advocate deferring preoperative biopsy in apparently resectable cases because histologic diagnosis is not required before surgery, although it is clearly required before embarking on neoadjuvant therapy or declaring a patient unresectable or having metastatic disease. Older literature suggested a benefit from treating patients already biopsied with low-dose (5 to 10 Gy) preoperative radiation in an attempt to decrease dissemination of tumor.⁴⁹ This is now very rarely done and would not be considered standard. Endoscopic retrograde cholangiopancreatography (ERCP) is extremely useful in the differential diagnosis of tumors of the pancreaticobiliary junction (Fig. 38-5).⁵⁰ Ampullary and duodenal cancers can be visualized and biopsied during ERCP; pancreatic duct stricture longer than 10 mm, especially if irregular, is more consistent with carcinoma than pancreatitis.⁵¹ Supportive serologic evidence of pancreatic cancer includes elevated levels of cancer antigen (CA) 19-9,⁵² carcinoembryonic antigen, pancreatic oncofetal antigen, tissue polypeptide antigen, and CA-125. A decrease in serial CA 19-9 levels has been found to correllate with survival; however, caution is warranted because of false elevations in the presence of benign biliary disease or obstruction or false negativity in individuals who are Lewis-a antigen negative. Unfortunately, no currently available marker is diagnostic or sufficiently sensitive or specific for screening purposes.⁵³ Following serial levels of these markers, however, is very useful, and baseline and post-treatment levels of CA 19-9 do correlate with tumor response.⁵⁴

FIGURE 38-4 • Computed tomographic scan showing large pancreatic mass; thin strip of barium indicates near-total duodenal obstruction. This underscores the need to have the treatment field cover the entire duodenal circumference in masses involving the pancreatic head. Note also that a significant volume of the right kidney would be in a field with an anterior-posterior–posterior-anterior arrangement.

FIGURE 38-5 • Endoscopic retrograde cholangiopancreatography of the same patient shown in Fig. 38-4. A, Note thin ragged contour of the pancreatic duct—stenosis of the length (see text) is usually indicative of neoplasm. B, A stent has successfully been placed in the pancreatic duct to relieve obstructive jaundice.

Following this, in patients without prior abdominal surgery, laparoscopy may be used to rule out small hepatic or peritoneal metastasis. Laparoscopic ultrasound in conjunction with peritoneal cytologic findings significantly decreases the nontherapeutic laparotomy rate as positive peritoneal cytologic findings indicate advanced disease. Investigators at MGH prospectively evaluated 88 patients referred for curative resection using MRI and CT, angiography, and laparoscopy for staging. They preoperatively identified 89% of patients who could not have curative resection. These patients are best spared an operation unless duodenal obstruction is present.⁴⁸ Excellent palliation for jaundice can be achieved with endoscopic placement of large stents.⁵⁵ Diagnostic workup and tumor-node-metastasis (TNM) classification are outlined in Table 38-2 and Table 38-3. It should be noted that the TNM staging criteria include data obtained both preoperatively and postoperatively and that the American Joint Committee on Cancer also recommends that the surgeon assess the completeness of resection as R0 for complete resection with negative margins, R1 for incomplete resection with microscopically positive margins, and R2 for incomplete tumor resction with gross residual tumor.⁵⁶ For practical purposes and to better understand the historical literature on the treatment of pancreatic cancer, much of it predating high-resolution imaging modalities, many of the clinical series have reported the conceptual classifications for pancreatic cancer patients as resectable, borderline resectable, locally advanced, and metastatic. This leads to confusion and difficulty in comparing results across institutions and trials.

Table 38-3 Evaluation of Pancreatic Mass

CA, Cancer antigen; ERCP, endoscopic retrograde cholangiopancreatography; GI, gastrointestinal; MRI, magnetic resonance imaging; PET, positron emission tomography.

Therapeutic Approaches

Options for treating any malignancy include surgery, radiation, and chemotherapy, either singly, sequentially, or in combination. The overall disappointing outcomes in the treatment of pancreatic cancer have involved all modalities in every order and combination possible. The standard therapeutic approach to pancreatic carcinoma has been surgical, and the standard operation for cancers in the pancreatic head has been the pancreaticoduodenectomy or Whipple procedure. Complete surgical resection offers the only possibility for cure, yet more than 80% of patients presenting with pancreatic cancer cannot be cured with resection.⁵⁷ Interestingly, there are no universally accepted criteria defining surgical resectability, although the driving consideration for the multidisciplinary team evaluating a potential surgical candidate is the likelihood of obtaining a complete resection with negative microscopic margins.^58,59 Tumors of the body and tail of the pancreas generally present late and thus usually are not resectable. Although the Whipple is the standard procedure, some controversy still surrounds the debate of the optimal surgerical technique. One controversy is the role of preoperative biliary drainage, with a 1999 MSKCC series⁶⁰ showing increased morbidity, including mortality, and a similar MD Anderson Cancer Center series⁶¹ showing no increase in sepsis or death. Thus most institutions selectively decompress the biliary tract in symptomatic patients or in those for whom surgery will be delayed. In most instances, this can be done by ERCP. CA-19-9 levels should be drawn after biliary decompression. Other surgical questions under investigation include the analysis of the optimal pancreatic anastamosis to minimize leaks and whether pylorus preservation offers a better quality of life and nutritional status. Total pancreatectomy has been advocated by some as a better operation because it removes more potentially involved tissue and because it avoids a pancreaticojejunal anastomosis, which is the source of significant surgical morbidity and mortality. However, it does result in exocrine insufficiency and brittle diabetes that is difficult to manage,⁶² and it confers no survival advantage. Fortner⁶³ advocated a regional pancreatectomy, which is an en bloc resection of the pancreas and surrounding tissue along with a segment of the mesenteric-portal vein and occasionally major arterial segments. This operation is less frequently performed in the United States because of high operative mortality (8% in Fortner’s series) and no clear evidence of survival benefit. Some groups in Japan combine the Whipple procedure with extensive lymph node dissection, claiming better results with no apparent increase in operative mortality.⁶⁴ The opposite approach has been taken by several groups in the United States^65,66 who believe there is no compromise of survival when preserving the pylorus in an attempt to avoid postgastrectomy symptoms. In the 1960s and 1970s, the pancreaticoduodenectomy was associated with a 20% to 25% in-hospital mortality rate.⁶⁷ By the 1980s and 1990s, surgical techniques and supportive care improved to the point at which this operation now carries only a 1% to 3% mortality rate in major centers.⁶⁸ With certainty, the pylorus-preserving pancreaticoduodenectomy improves gastrointestinal function and does not appear to compromise further management. The dramatic improvements in surgical techniques have made it possible to conduct an ongoing prospective randomized trial with intraoperative randomization between a standard operation and one with extended retroperitoneal lymph node dissection. What is clear from the surgical literature is that, if tumor is knowingly left behind, survival after resection is not much better than after bypass alone.⁵³ This has led to the question of appropriate surgical management of patients with positive lymph nodes at frozen section. Yeo and colleagues⁶⁸ have shown that long-term survival out to 5 years is possible in both lymph node–positive and lymph node–negative tumors as long as the patient is otherwise technically suitable for resection. An additional important question answered recently involves determination of the best management of the elderly patient. Sohn and colleagues,⁶⁹ Magistrelli and colleagues,⁷⁰ and Balcom and colleagues⁷¹ clearly showed that it is appropriate to perform aggressive surgery in appropriately selected patients into their seventh and eighth decades of life. Several other important points from the surgical literature are worth noting. These involve the importance of both a multidisciplinary approach as well as the importance of an experienced team. The importance of an experienced team is further underscored by series from Robinson et al.⁷² and from Sohn et al.⁷³ at two different institutions, showing the benefit of reoperation at an experienced center: 67% of patients, many deemed unresectable elsewhere, were in fact successfully resected without compromise of their overall outcome. Other series have clearly shown the benefit of decreased in-hospital morbidity and mortality rates as well as survival in those centers with a high volume of cases.^74,75

Multiple large series have shown that only 5% to 30% of patients have resectable tumors.^64,76–78 Remaining patients require some form of palliation for pain, duodenal obstruction, or jaundice. Obstructive jaundice can be relieved percutaneously, albeit with complications in up to 35% in some series,⁵³ or with stents that have more than an 85% success rate and procedure-related mortality of only 1% to 2%.⁷⁹ This represents a significant improvement over surgical morbidity. Two randomized trials^80,81 have shown equal survival with both surgical bypass and endoscopic stent placement. In addition to analgesics, percutaneous neurolysis of the celiac ganglion and radiation therapy offer good palliation of pain.

In addition to various surgical approaches, many other avenues have been pursued to improve the outcome for patients with pancreatic cancer. These include not only chemotherapy and preoperative and postoperative external-beam radiation therapy (EBRT) but also intraoperative radiation therapy (IORT), brachytherapy with I-125 and Au-198 or Pd-103, and treatment with heavy particles.

Another strategy to treat in patients with resectable pancreatic adenocarcinomas is preoperative treatment with gemcitabine and radiotherapy. The rationale for this approach is as follows: (1) early systemic treatment with gemcitabine addresses risk of micrometastatic disease when tumor burden is lowest, (2) gemcitabine is a potent radiosensitizer, and (3) restaging studies performed after neoadjuvant therapy selects out those patients who are not good surgical candidates because of rapid disease progression. In a phase II study of 86 patients, this approach resulted in a resection rate of 74% having a median survival rate of 34 months. The median survival rate in the entire group of 86 was 22.7 months. The 5-year survival rate in those patients able to undergo successful surgical resection was 36% versus 0% in those who did not undergo surgery.⁸²

For patients with resectable disease, the role of adjuvant chemoradiotherapy was established by the Gastrointestinal Tumor Study Group (GITSG) in 1985, when it showed that patients receiving adjuvant chemoradiation experienced a doubling of survival time.⁸³ Small series using neoadjuvant treatment were reported in the 1980s, but that approach did not attract general interest until improved outcomes in surgical series were reported.^84,85 Historically 5-fluorouracil (5-FU) was the chemotherapeutic agent of choice (Table 38-4) until Burris et al.⁸⁶ showed an improvement in survival with gemcitabine over 5-FU as first-line therapy for patients with advanced pancreatic cancer. Subsequently, gemcitabine and 5-FU in varying doses and dose rates have been combined with and without radiation with varying results.^87–89

Gemcitabine, now the most widely used chemotherapy in pancreatic cancer, has shown significant promise; but combination studies with other chemotherapies and targeted agents have been largely dissappointing. Gemcitabine has been combined with radiation therapy with modest success. However, because of the significant radiosensitizing properties of gemcitabine, reduced radiation dose and field size is necessary if full-dose gemcitabine is administered. Investigators from the University of Michigan⁹⁰ reported a median overall survival (OS) of 11.2 months in patients (n = 74) with locally advanced pancreatic cancer treated with full-dose gemcitabine (1000 mg/m², weekly) and concurrent radiotherapy (36 Gy in 2.4-Gy fractions). Using three-dimensional (3-D) conformal techniques, they treated the primary tumor with a 1-cm margin. Despite not treating the regional lymph nodes, only 5% of patients failed regionally, suggesting that treatment of the regional lymph nodes may be omitted. Similarly, Chang et al.⁹¹ reported that 0 of 77 patients developed an isolated nodal failure after stereotactic radiotherapy targeting only the primary tumor. Overall, there were nine nodal recurrences, but all were in the context of progressive disease at other sites.

Other strategies for improving the effectiveness of radiotherapy for pancreatic cancer include intensity-modulated radiotherapy (IMRT),⁹² image-guided radiotherapy (IGRT),⁹³ and stereotactic radiotherapy.^94,95

The major difficulty in treatment of pancreatic cancer is not the radioresistance of the tumor, but radiation dose limitations imposed by surrounding normal organs (see Fig. 38-1B) and the frequent presence of occult distant metastases. Attempts to overcome this problem have included the use of intraoperative electrons or implantation of the tumor either directly or percutaneously. Obviously, the advantage offered by IORT is precise field localization and the abilities to shield sensitive normal structures and to deliver a relatively high dose quickly. High-dose-rate brachytherapy has the advantage of reduced radiation exposure to medical personnel and the potential to overcome limitations of attempting to control tumors with a relatively fast doubling time with a low-dose-rate source.

Discouraging survival statistics, a 50% to 76%^37,38 locoregional failure even after “curative” resection, and the proven survival benefit of adjuvant treatment^83,96 make postoperative radiation with concurrent chemotherapy justifiable in virtually all patients who have had a potentially curative resection of pancreatic cancer. Patients with localized but unresectable cancer can expect doubling of survival with EBRT and concurrent 5-FU.^97–99 They should be offered treatment unless poor clinical status precludes treatment.

Simulation

With the use of multiple fields and high-energy photons, doses of 60 to 65 Gy can be delivered safely.¹⁰⁰ The addition of concurrent chemotherapy typically requires dose reduction to avoid excessive toxicity. With the widespread development and availability of IMRT techniques, increased dose intensity to the tumor is possible while minimizing radiation dose to adjacent normal tissues. Furthermore, reduced tumor margins also are possible with the application of IGRT techniques.

In patients undergoing an operation, the surgeon should outline the tumor with small vascular clips if, intraoperatively, the tumor is determined to be unresectable. In patients with resectable disease, surgical clips are useful to help define the tumor bed and anastamotic sites. If used sparingly, CT artifacts are not significant. Titanium clips produce less interference on CT, but they are often not seen on standard x-ray films. The target volume, which includes the tumor and nodal groups at risk, is further delineated by information obtained on preoperative and postoperative CT, preoperative barium studies, ERCP, angiography, ultrasound, and the findings at operation. It is particularly important that the radiation oncologist understand which surgical procedure was performed so that this information can be incorporated into treatment planning. Given the overall poor outcome in patients with pancreatic cancer, one may intuitively question the efficacy of prophylactic treatment of nodal groups in patients with gross disease. This issue has never been addressed in a formal or prospective fashion, but the trend in the field is toward reduced radiation fields, particularly with the development of gemcitabine-based chemotherapy.

The patient is placed in the supine position with arms comfortably on the upper chest or above the head. Renal contrast is given and an initial anterior-posterior (AP) and lateral films are taken to establish the position of the kidneys. Subsequently, oral contrast is given to visualize the stomach and duodenal C-loop, which will localize the position of the head of the pancreas in those patients who have not undergone a Whipple procedure.

Anterior-Posterior–Posterior-Anterior Field: Lateral Borders

For lesions of the head of the pancreas, treatment of nodal groups, if indicated, should include the pancreaticoduodenal, suprapancreatic, celiac, and porta hepatis lymph nodes. Lesions in the head of the pancreas frequently invade the medial wall of the duodenum, and therefore the entire involved duodenal wall should be covered for lesions. In the direction of the pancreatic body, the treatment volume should extend to include a 2- to 3-cm margin beyond gross disease. Care must be taken to keep doses within renal tolerance. Often more than 50% of the right kidney is within the field; thus for lesions of the pancreatic head, at least two-thirds of the left kidney should be excluded from the anterior-posterior–posterior-anterior (AP–PA) portion of the treatment field. When the primary tumor is located in the pancreatic head, the risk of nodal involvement in this area is quite low (see Fig. 38-2B).³⁵ The risk of renal injury is very low if half of the renal mass receives no more than 18 Gy.⁴⁹

For lesions in the pancreatic body or tail, the target volume includes the tumor with a 2- to 3-cm margin, pancreaticoduodenal and porta hepatis nodes, lateral suprapancreatic nodes, and nodes of the splenic hilum. It is not necessary to include the whole duodenal loop in the treatment field, and thus most of the right kidney can be spared. Often 50% of the left kidney is in the field when the more lateral nodes are covered (Fig. 38-6).

FIGURE 38-6 • A and B, Simulation films of the patient shown in Fig. 38-4. See text for discussion. Care is taken to include all relevant at-risk regional nodes and to spare as much of one kidney as possible. On the lateral fields, a block should be placed to shield the spinal cord but not block at-risk para-aortic lymph nodes.

Anterior-Posterior–Posterior-Anterior Field: Superior and Inferior Borders

Kao and colleagues¹⁰¹ demonstrated significant variability in the location of the celiac axis and superior mesenteric nodes (Fig. 38-7), which emphasizes the need for individualized treatment planning when attempting to cover nodal groups associated with these vessels. For pancreatic head lesions, to ensure adequate coverage on the celiac nodes, the superior border should be at the T10/T11 level and the lower border at the L3/L4 level, depending on the preoperative staging studies. Guidelines are identical for lesions of the body and tail, although the superior margin occasionally is higher because of the location of the primary tumor.

FIGURE 38-7 • Graphic representation of the specific locations of the origin of the celiac axis (A) and superior mesenteric arteries (B). This underscores the need for individualized treatment planning when covering these nodal groups.

(From Kao GD, Whittington R, Coia L: The anatomy of the celiac axis and superior mesenteric artery and its significance in radiation therapy. Int J Oncol Biol Phys 25:131, 1993. Reprinted by permission of Elsevier Science, Inc.)

Lateral Field Borders

Superior and inferior margins remain the same. Care must be taken to be aware of the location of the kidneys and to take this into account when assigning weighting to the lateral fields. The anterior margin is placed 1.5 to 2 cm beyond gross disease as defined on preoperative CT scan. For the posterior margin, the spinal cord is blocked, but at least 1.5 to 2 cm of the anterior portion of the vertebral body is in the field. This is to allow margin on the para-aortic nodes, which are at risk from posterior extension of tumor.

The lateral field contribution is usually limited to 15 to 18 Gy because of inclusion of a large portion of the kidney and liver in the AP–PA fields. If treating to high doses without chemotherapy, field reduction should be made at 50.4 Gy to help diminish the probability of injury to the stomach and small bowel. Otherwise, field reductions should be made at 45 Gy. Boost fields typically dose an additional 9 Gy.

Significantly larger fields encompassing the liver have been studied by some,¹⁰² but are not considered standard.

CT-based radiation treatment planning is widely available and should be considered the standard of care (Fig. 38-8). It is possible to determine complete anatomic and dose information for the entire tumor volume and surrounding normal tissues (Fig. 38-9A, B). An approach using tight conformal fields tailored to each patient is highly desired when treating aggressively with the newer chemotherapeutic agents that have higher toxicity than 5-FU. Construction of dose-volume histograms shows the volume of normal tissue in the treatment field (see Fig. 38-9C). Use of this information to assess the probability of normal tissue complications is under active investigation. The initial simulation is accomplished as described earlier with the patient placed in an immobilization device such as an alpha-cradle. The patient is then CT scanned in the immobilized treatment position on a flat CT tabletop with the CT alignment lasers positioned on the patient’s skin marks so that the simulation position can be exactly duplicated. Alternately, a dedicated CT simulator may be used. CT slices typically are obtained at 3- to 5-mm intervals, although even higher resolution imaging using biphasic high-speed helical CT scanning at 1.25-mm intervals is desirable for optimal definition of the tumor and regional anatomy. The clinician then outlines the target volume on each CT slice, which includes tumor, areas at risk for microscopic involvement, and an additional 10-mm margin to allow for setup error and patient motion. If daily IGRT techniques are used, the margin expansion can be reduced. Other critical organs, such as the kidneys and spinal cord, also are contoured, and these as well as the target volume and bony landmarks are reconstructed in three dimensions and can be displayed using a beam’s-eye view (BEV) technique. Beam shaping is then accomplished in this mode, ensuring adequate coverage of the target volume with minimal inclusion of normal tissue. Once the optimal beam arrangement has been determined, the CT image set is used to generate a BEV template at the same magnification as the initial simulation film for each treatment field. Often the isocenter may have been moved during the treatment planning process. At verification, the patient is set up again with the initial parameters, and changes are then made as determined by the 3-D plan. The new BEV template established by the 3-D plan is overlaid onto the verification simulation films (obtained for each field). The template can then be used to fabricate custom blocks (Fig. 38-10).

FIGURE 38-8 • Axial (A), coronal (B), and sagittal (C) color-wash isodose distributions of three-field plan with lateral wedges. The tumor bed volume is covered in the 95% isodose line with a hot spot of 104%. The anterior field is weighted 40% whereas the laterals have weighting of 30% each. Block margins for this T₁N₀M₀ tumor are 1.0 to 1.5 cm.

FIGURE 38-9 • Beam’s-eye views, anterior (A) and lateral (B), of the plan from Fig. 37-8 showing a color wash of the target and significant surrounding structures. These same data are represented graphically in a dose-volume histogram (C).

FIGURE 38-10 • Anterior (A) and lateral (B) simulation and anterior (C) and lateral (D) port films for the three-dimensional treatment plan shown in Fig. 38-8.

Intensity-Modulated Radiotherapy

Ben-Josef and colleagues⁹² described using an IMRT approach to treat locally advanced and resected pancreatic cancer with concurrent capecitabine. Using this technique, the primary tumor (or tumor bed if treating postoperatively) was treated with 54 Gy and the lymph nodes with 45 Gy. Acceptable toxicity occurred if the following dose limitations were observed: 50% of each kidney below 20 Gy, 67% of liver below 35 Gy, 90% of small bowel below 45 Gy, 90% of stomach below 45 Gy, and 90% of spinal cord below 45 Gy (max 50 Gy). Milano et al.¹⁰³ also reported acceptable toxicity using IMRT for pancreatic and bile duct cancer. The majority of these patients also received 5-FU. These investigators recommended the following dose guidelines: 10% to 20% of the left kidney more than 12 Gy (max 30 Gy), 25% to 30% of the right kidney more than 20 Gy, 50% to 60% of the liver more than 8 Gy (max 44 Gy), and spinal cord less than 45 Gy. When treating pancreatic tail lesions, the dose constraints for the right kidney were tightened and the dose constraints for the left kidney were relaxed. Small bowel constraints were not considered in planning. In this series, unresectable tumors typically were treated to 59.4 Gy and regional lymphatics to 45 Gy.

Image-Guided Radiotherapy

Pancreatic tumors can move significantly in the course of normal respiration. The motion is greatest in the inferior-superior direction, but significant motion also exists in the AP and left-right directions. In one observational study of liver and pancreatic tumors with implanted fiducial clips, the average range of motion as determined by quantitative fluoroscopic analysis in the craniocaudal direction was 7.4 mm.¹⁰⁴ The investigators concluded that tumor motion of this magnitude could significantly affect IMRT dose distributions. More recently, Minn and colleagues¹⁰⁵ reported that, in a group of patients with pancreatic cancer, the maximum tumor motion that occurred during treatment was greatest in the superior-inferior direction (mean: 6.5 mm, range: 0.5-12.7 mm) but also occurred in the AP (mean: 2.9 mm, range: 0.6-5.5 mm) and left-right (mean: 2.4 mm, range: 0.4-9.4 mm) directions. Dynamic MRI and ultrasound studies suggest that pancreatic tumor motion occurs as much as 20 to 35 mm throughout the respiratory cycle.^106,107

If smaller treatment margins are used, tumor motion compensation strategies such as respiratory gating should be considered. The most direct method to ensure accuracy is to use daily kV or fluoroscopic imaging of fiducial seeds implanted into the pancreatic tumor. Daily shifts are then made to align these seeds in accordance with the planning scans. Advances and refinements in IGRT techniques hold great promise to reduce normal tissue toxicity and to increase radiation dose intensity to the tumor.

Intraoperative Radiation Therapy

Preoperative CT is used to do the initial planning with regard to required cone sizes, shapes, and angulations that will be needed. Anticipated electron energy also is determined. Malleable lead shields thick enough to reduce the electron dose by 90% are sterilized. These will be used as secondary collimation to block normal structures within the treatment field. At the time of surgery, maximal exposure of the tumor should be obtained, and the tumor bed should be outlined by surgical metallic clips. An intraoperative ultrasound supplements the CT information in determining the electron energy to be used. The IORT field should encompass the entire tumor bed as outlined with metallic clips plus a small margin to allow for dose falloff at the field edge. At the same time, the cone size should be chosen to spare as much normal tissue as possible. At least two-thirds to three-fourths of one functional kidney should be excluded from the field; this can be accomplished with the use of the previously mentioned malleable lead shields. The most common dose that has been used, which is also the current Radiation Therapy Oncology Group (RTOG) recommendation, is 15 Gy.

Chemotherapy Alone

A multitude of published series have shown improved disease-free survival (DFS) and OS with the addition of 5-FU to radiation. Various doses and schedules have been used; however, no prospective randomized study has clearly shown superiority of one schedule over another. Chemotherapy alone generally is reserved for patients with metastatic disease or poor-performance status patients with locally advanced disease. The only drugs currently approved by the U.S. Food and Drug Administration (FDA) for the management of patients with pancreatic ductal adenocarcinoma are gemcitabine and erlotinib. Gemcitabine initially was studied in a randomized trial and compared with weekly bolus 5-FU.⁸⁶ The primary endpoint was clinical benefit as defined by performance status, narcotic use, and weight loss. In this pivotal trial, gemcitabine was found to be superior to 5-FU in clinical benefit. In addition, this trial showed a small but significant survival advantage. Subsequently, there have been many studies of gemcitabine combinations with both traditional chemotherapy drugs such as cisplatin, oxaliplatin, capecitabine, and irinotecan and new targeted therapeutics such as a farnesyl transferase inhibitor, bevacizumab, cetuximab, and erlotinib. Only the combination of gemcitabine and erlotinib¹⁰⁸ has shown statistically significant improvement in OS compared with gemcitabine alone. However, because the OS benefit in this trial was small (hazard ratio 0.82), this combination has not been adopted as a standard of care, but is certainly an option for patients. Several other drugs have shown improvement in objective response rate or progression-free survival in randomized clinical trials.^109–112 These combinations have included gemcitabine plus cisplatin,¹⁰⁹ oxaliplatin,¹¹⁰ capecitabine,¹¹¹ and bevacizumab.¹¹² Improvement in progression-free survival with bevacizumab was seen only in combination with gemcitabine and erlotinib,¹¹³ not with gemcitabine alone.¹¹³ Current National Comprehensive Cancer Network guidelines suggest that chemotherapy combinations that include erlotinib, cisplatin, oxaliplatin, and capecitabine are acceptable options in patients with advanced disease who have a good performance status.¹¹⁴

Critical Normal Tissues

Critical normal tissues in a typical field treating pancreatic carcinoma include the liver, small bowel, stomach, kidney, and pancreas. The fact that the pancreas is more radioresistant than surrounding organs, coupled with overall poor long-term survival, has made reports of radiation injury to the pancreas exceedingly scarce in the literature. Fajardo and Berthrong¹¹⁵ describe radiation effects to be primarily stromal fibrosis, myointimal proliferation of the arteries, ductal stenosis, and parenchymal atrophy that is similar to the appearance of chronic pancreatitis without the active inflammatory component. Islet cells appear to be more resistant than the exocrine tissue. Autopsy studies in patients treated with helium ions¹¹⁶ also have shown prominent marked interstitial fibrosis and vascular sclerosis.

Outcome

Treatment for Localized but Unresectable Tumors

Optimal therapy for patients with locally advanced pancreatic carcinoma remains controversial largely because of the great clinical heterogeneity within that population. Median survival for localized unresectable cancer of the pancreas treated with bypass alone is 4 to 6 months.^{29,30,141–143} EBRT alone does not offer significant improvement.^98,99,144 Hyperfractionated radiation, 1.2 Gy twice daily, was also disappointing, although the total dose of 50.4 Gy may have been too low to expect significant benefit.¹⁴⁵ Particle therapy to include pions,¹⁴⁶ neutrons,^147–150 and helium ions^151,152 has been disappointing (Table 38-6). Neutron therapy hoped to take advantage of a higher relative biologic effectiveness (2.6 to 3.0), and helium ion therapy was used to take advantage of improved dose localization properties. Even though initial data with helium appeared to be very promising¹⁵¹ and autopsy studies did confirm tumor kill with minimum effect on adjacent local tissue,¹¹⁶ a randomized prospective trial comparing helium ions with conventional radiation failed to show a survival benefit in spite of improved local control in the helium arm.¹⁵³ The photon arm was split-course and the helium continuous; failures were due largely to distant metastases. EBRT combined with chemotherapy, however, has been shown to confer significant survival advantage over either EBRT or chemotherapy alone (Table 38-7). Although Klaasen et al.¹⁵⁴ found no difference with combined modality treatment, the GITSG in a prospective study of 194 patients¹⁴⁴ did find that the addition of 5-FU to EBRT almost doubled the mean survival time (from 5.5 to 10 months) and that at 1 year 40% of the combined modality group was alive versus only 10% of those treated with radiation alone. Other investigators have since duplicated these findings.^98,99 The cooperative groups also have explored the newer chemotherapeutic agents. An Eastern Cooperative Oncology Group trial using protracted infusion 5-FU, gemcitabine, and EBRT showed unexpected early toxicity,¹⁵⁵ necessitating early study closure. This underscored earlier studies that suggested that relatively high doses of radiation, large fields including lymph nodes at risk, and gemcitabine were too toxic. Similar experience of high toxicity suggesting a very narrow therapeutic index with gemcitabine chemoradiotherapy was demonstrated by researchers at MD Anderson Cancer Center.¹⁵⁶

An alternate approach initially was investigated at the University of Michigan, consisting of a standard dose of gemcitabine (1000 mg/m²) weekly along with escalating doses of radiation.¹²⁶ Radiation fields were tightly 3-D planned, covering only gross tumor volume. Doses ranged from 24 to 42 Gy delivered in 3 weeks with 1.6- to 2.8-Gy fractions. This study included 37 unresectable or incompletely resected patients, 14 of whom had confirmed or suspected metastatic disease. Median survival of 11.6 months and patterns of failure suggested that field-size reductions were not harmful and led to this approach being widely adopted. Subsequently, another approach—systemic chemotherapy prior to chemoradiation in the setting of locally advanced disease—was used. Both retrospective analyses¹⁵⁷ and prospective studies^158,159 suggest that treating patients with locally advanced disease with systemic chemotherapy first helps to establish overall disease control and selects for patients with locally advanced disease who are more likely to benefit from subsequent chemoradiation therapy. A significant number of patients who appear to have locally advanced but nonmetastatic cancer quickly fail and do not benefit from the locoregional treatment approach. The therapeutic strategy used by Huguet et al.¹⁵⁷ in the GERCOR (French Oncology Research Group, Paris, Fance) II and III studies (n = 181 patients) was to administer initial chemotherapy for 3 months and thus select out those patients who would then benefit from the addition of further chemotherapy and radiation. In this study, tumor progressed in 29% of patients during the first 3 months and their median OS was only 4.5 months. Patients without tumor progression during the first 3 months had a median survival of 13.1 months. As a result, the patients who then received the course of chemotherapy followed by planned chemoradiotherapy had a significantly better OS of 15 months versus 11 months if treated with chemotherapy alone. Although registered on the trial in a prospective manner, Huguet’s study is a retrospective analysis of the 128 patients (from the original 181) who were without disease progression at 3 months, which leads to some criticism of the introduction of bias, mandating a future prospective trial to consider whether patients with locally advanced, nonmetastatic disease should all be offered up-front chemotherapy prior to radiation therapy. The GERCOR and Arbeitsgemeinschaft Internistische Onkologie groups are now conducting this phase III trial.