Chapter 21: Pancreatic Cancer

When clinicians use the term pancreatic cancer, they refer to adenocarcinoma of the pancreas, one of the most challenging malignancies facing oncologists today. This disease is characterized by significant morbidity and poor prognosis.

At the University of Texas MD Anderson Cancer Center (MDACC), we manage patients with pancreatic cancer with a multidisciplinary team and view palliation as the primary goal. However, for patients with potentially resectable disease, we take an aggressive approach whenever appropriate. In the setting of advanced disease, cure is not possible, but as our understanding of carcinogenesis, invasion, and metastasis expands, more effective therapeutic strategies are expected to emerge. This chapter reviews our current knowledge about pancreatic cancer, including its epidemiology, risk factors, molecular biology, diagnosis and staging, and clinical strategies for therapy.

Pancreatic cancer, the most common pancreatic neoplasm, is an aggressive and often rapidly fatal malignancy. In the United States, it represents 2% of all cancer cases but accounts for 5% of all cancer deaths (¹). Currently, it is the fourth leading cause of cancer death, ranking behind lung, colorectal, and breast cancer. While evidence suggests marginal improvements in 5-year survival rates over the last 25 years (2% in 1974-1976, 3% in 1983-1985, and 4% in 1992-1997), life expectancy remains short and is generally measured in months (²). By 2030, deaths due to pancreas cancer are projected to increase dramatically and will become the second leading cause of cancer-related death (³). Significant improvements in survival have been hampered by a number of factors, including inefficient screening strategies, technically challenging and often debilitating surgery, and minimally effective chemotherapy and radiotherapy.

Pancreatic cancer is a dynamic disease, and sudden changes in clinical status occur frequently. Patients may rapidly develop worsening pain, biliary obstruction, or stent occlusion with cholangitis, thromboembolism, peritoneal carcinomatosis, or intractable ascites. Any of these problems may preclude the timely delivery of cytotoxic therapy and limit survival. Therefore, most efforts should focus on symptom control, but for patients with adequate performance status (PS), treatment is encouraged.

There are approximately 45,000 new cases of pancreatic cancer each year in the United States and 330,000 cases worldwide. Incidence rates are highest in industrialized societies and Western countries. Of note, the risk of pancreatic cancer among African Americans, in whom pancreatic cancer mortality rates are higher than most other ethnic groups in the United States, is considerably higher than the rates for African blacks (⁴). These observations implicate environmental factors conspiring with genetic background as causes of the increased risk.

The risk of developing pancreatic cancer is low in the first three to four decades of life but increases sharply after the age of 50. Average age at the time of diagnosis is 72 years. Pancreatic cancer is uncommon in patients under 40. In the past, pancreatic cancer occurred more frequently in men, but now the disease is becoming more common in women, probably secondary to the increased use of tobacco by women.

Surprisingly, relatively little is known about the risk factors for the development of pancreatic cancer. Table 21-1 summarizes genetic and environmental factors associated with an increased risk.

Tobacco

Aside from age, the only consistently reported risk factor for pancreatic cancer is cigarette smoking. Cigarette smoking is estimated to account for roughly 30% of pancreatic cancer mortality (⁵). Experimental models have demonstrated that nitrosamines found in tobacco smoke are carcinogenic for the pancreas. Research performed at MDACC has shown that smoking-related aromatic DNA adducts and other types of DNA damage may be critical in carcinogenesis.

Diabetes Mellitus

Diabetes has been implicated both as an early manifestation of pancreatic cancer and as a predisposing factor. A meta-analysis published between 1975 and 1994 showed that pancreatic cancer occurred with increased frequency in patients with long-standing diabetes (diagnosed at least 5 years prior to the diagnosis or death due to pancreatic cancer) (⁶). It is believed that insulin resistance and secondary hyperinsulinism may be involved in pancreatic carcinogenesis.

Chronic Pancreatitis

Early clinical studies have suggested an association between chronic pancreatitis and pancreatic cancer. In a study of 715 patients with chronic pancreatitis diagnosed between 1971 and 1995, there was a 13- to 18-fold increase in pancreatic cancer rates (⁷).

Studies have suggested that hereditary pancreatitis, in particular, may affect the risk. Lowenfels and Maisonneuve studied 246 patients with hereditary pancreatitis from 10 countries. The estimated cumulative risk of pancreatic cancer by age 70 was approximately 40%, with a mean age at diagnosis of 57 years (⁸). Molecular data strongly suggested that mutations in the trypsinogen gene PRSS1 play an important role in hereditary and possibly acquired forms of pancreatitis (⁹).

Diet

Positive associations have been discovered between pancreatic cancer and meat and carbohydrate intake. However, there is no consensus on the contribution of dietary fat. Epidemiologic studies of pancreatic cancer have shown a protective role for high fruit and vegetable intake. This effect may be related to dietary folate and other methyl donor groups.

Body Mass Index

Epidemiologic studies have also implicated a high body mass index as increasing risk for pancreatic cancer. This may be explained by relative hyperinsulinemia thought to promote pancreatic carcinogenesis (¹⁰).

Familial Pancreatic Cancer and Other Genetic Syndromes

Patients with pancreatic cancer who have two first-degree relatives with a history of pancreatic cancer are defined as having familial pancreatic cancer. The estimated relative risk for other family members is increased 10- to 20-fold over the general population. In addition, hereditary nonpolyposis colorectal cancer, ataxia-telangiectasia, Peutz-Jeghers syndrome, familial breast and ovarian cancer, and familial atypical multiple-mole melanoma are all associated with increased risk of pancreatic cancer (¹¹).

The familial breast and ovarian cancer syndrome, associated with mutations in the BRCA genes, accounts for 17% of cases (¹²). The BRCA1 and BRCA2 proteins are key components of homologous recombination through the repair of DNA double-strand breaks. Cells deficient in these proteins demonstrate genomic instability and a tendency toward malignant transformation (¹³). The poly(ADP-ribose) polymerase (PARP) enzyme is integral for single-stranded DNA repair, and inhibition results in DNA breaks that require intact BRCA proteins for repair. In the presence of mutant BRCA proteins, PARP inhibition results in synthetic lethality (¹⁴).

Preclinical work has demonstrated the sensitivity of BRCA mutated pancreatic cancer cell lines to cross-linking chemotherapeutic agents, such as mitomycin-c and cisplatin (¹⁵). This finding has translated clinically as studies have shown superior overall survival and response in patients with advanced pancreatic cancer with BRCA mutations treated with platinum (¹³). Furthermore, studies with PARP inhibitors have shown promise either alone or in combination with cytotoxic chemotherapy (¹⁶). Currently, an ongoing study is investigating the efficacy of a single-agent PARP inhibitor, rucaparib, in patients with advanced pancreatic cancer with known BRCA mutations.

Occupational Exposures

Exposure to carcinogens in the workplace has been implicated in pancreatic cancer, but with the possible exception of formaldehyde, the available evidence is insufficient to identify any specific exposure to increase the risk.

Prior Gastrointestinal Surgery

Surgical procedures such as gastrectomy and cholecystectomy have been reported to increase risk and are possibly linked to elevated levels of cholecystokinin and hypergastrinemia. However, other studies have not demonstrated such an association (¹⁷).

The molecular events leading to pancreatic cancer have not been fully elucidated, but mutations in a few oncogenes and tumor suppressors appear critical for carcinogenesis. Investigators from Johns Hopkins performed a comprehensive genomic analysis of pancreatic cancer reported in the journal Science. They noted an average of 63 alterations, the majority of which were point mutations. These alterations defined a core set of 12 cellular signaling pathways and processes that were altered in the majority of pancreatic cancers. These included the K-ras, wnt/notch, hedgehog, TGF-β, integrin, and JNK signaling pathways (¹⁸). Recently, Waddell et al characterized pancreatic cancer into four subtypes termed stable, locally rearranged, scattered, and unstable (¹⁹).

Oncogene Mutation: The ras Oncogene

Pancreatic cancer has the highest frequency of K-ras mutation among all human cancers. Greater than 85% of cases have an activating point mutation in the K-ras gene, most commonly a G-to-T transversion at codon 12 (²⁰). This leads to constitutive activation of RAS, leading to growth-promoting signal cascades via the mitogen-activated protein (MAP) kinase pathway. Preclinical models have implicated this mutation as a very early event in carcinogenesis. The ras oncogene mutations were also retrospectively seen in pancreatic juice collected 3.5 years prior to a patient’s diagnosis of pancreatic cancer (²¹).

Tumor Suppressor Gene Mutation and Inactivation

The p16 tumor suppressor gene is inactivated in approximately 95% of pancreatic cancers; this typically occurs later in carcinogenesis. The second most frequently inactivated tumor suppressor gene, p53, located on chromosome 17p, also appears to be a late event in tumorigenesis. The DPC4 gene (SMAD4) is inactivated in 55% of pancreatic adenocarcinomas and, like p53, is a relatively late event in tumorigenesis. In a comprehensive mutational analysis of 42 pancreatic ductal cancers, Rozenblum and colleagues found individual mutational frequencies of tumor suppressor genes p16, p53, DPC4, and BRCA2 were 82%, 76%, 53%, and 10%, respectively (²¹).

The Multistep Sequence of Pancreatic Carcinogenesis

Current data suggest a temporal sequence of molecular events in pancreatic carcinogenesis leading from an “adenomatous” or proliferative ductal cell phenotype to pancreatic intraepithelial neoplasia (Pan-IN) and finally to invasive cancer. This theory has been supported by the recognition of noninvasive proliferative ductal lesions (Pan-IN1) that demonstrate mutations commonly found in pancreatic cancer. Growing evidence suggests that the gradual accumulation of genetic and biochemical alterations in early lesions causes progression to higher levels of dysplasia (Pan-IN2 and Pan-IN3) and ultimately to cancer.

Other Molecular Events in Pancreatic Carcinogenesis, Invasion, and Metastasis

The overexpression of epidermal growth factor receptor (EGFR), vascular endothelial growth factor (VEGF), matrix metalloproteinases, cyclooxygenase 2 (COX-2), hedgehog signaling, and insulinlike growth factor type 1 (IGF-1) pathways have been implicated in pancreatic cancer. Studies conducted at MDACC have also demonstrated that the nuclear transcription factor–κB (NF-κB) is commonly activated in pancreatic cancer (²²).

Hedgehog Signaling Pathway

Hedgehog signaling is an essential pathway during embryogenesis of the normal pancreas, and its dysregulation has been reported in precancerous PanIN-1 and -2 as well as in primary and metastatic pancreatic adenocarcinomas. Inhibition of hedgehog signaling by cyclopamine-induced apoptosis and blocked proliferation in vitro and in vivo (²³). Although debatable, hedgehog inhibition also depletes tumor-associated stroma and may play a role in the delivery of chemotherapeutic agents like gemcitabine (²⁴). Clinical trials of hedgehog inhibitors have failed to create a meaningful impact in this disease. Recent data indicated that stromal depletion may actually be deleterious, suggesting a critical need for accurate interpretation of preclinical data before incorporation into the clinical trial setting (²⁵).

Insulinlike Growth Factor Type 1 Pathway

Insulinlike growth factor type 1 upregulates cell proliferation and invasiveness through activation of the phosphoinositide 3-kinase (PI3K)/Akt signaling pathway and downregulates the tumor suppressor chromosome 10 (PTEN). Akt mediates the gemcitabine and erlotinib resistance mechanisms. At MDACC, we recently concluded a clinical trial with dalotuzumab (MK-0646), which targets the IGF-1 pathway.

Targeted therapy has been a major focus of clinical research, taking priority over the study of more conventional cytotoxic agents. Preliminary studies with agents designed to abrogate RAS function have been disappointing (²⁶). Likewise, administration of inhibitors of matrix metalloproteinases has failed to demonstrate a meaningful clinical impact (²⁷). Recent studies of EGFR, VEGF, NF-kB, or COX-2 inhibitors are discussed in further detail in this chapter. Immune therapies may have a significant impact as well.

Pancreatic acinar cells account for approximately 80% of the cell number and volume of the gland, with islet cells accounting for 1% to 2%. The ductal system is made up of single-layer, cuboidal epithelial cells comprising 10% to 15% of the gland’s structure, with a sparse interlacing network of blood vessels, lymphatics, nerves, and collagenous stroma. In carcinoma, this architecture is markedly altered: The predominant histologic feature is a dense collagenous stroma with atrophic acini, remarkably preserved islet cell clusters, and a slight-to-moderate increase in the number of normal-appearing and cancerous ducts (Fig. 21-1). The diagnosis of ductal adenocarcinoma rests on the identification of mitoses, nuclear and cellular pleomorphism, discontinuity of ductal epithelium, and evidence of perineural, vascular, or lymphatic invasion.

Almost all malignant neoplasms of pancreatic origin (95%) arise from the exocrine portion of the gland. Tumors arising from the islets of Langerhans (endocrine) cells are much more infrequent, and primary nonepithelial tumors (eg, lymphomas or sarcomas) are extremely rare. The histologic classification of exocrine pancreatic neoplasms is presented in Table 21-2.

The clinical presentation of pancreatic cancer is primarily dependent on the location of the tumor within the pancreas. The majority (85%) develop within the pancreatic head. About 10% are located in the pancreatic body and 5% in the tail. Nonspecific, poorly localized, epigastric or back pain is the most common initial presentation. It is usually caused by invasion or compression of the celiac, splanchnic, or mesenteric plexi. Tumors in the head or neck typically cause pain in the epigastric area or in the right upper quadrant of the abdomen. Cancers of the body may cause unremitting, severe back pain, and tumors in the pancreatic tail are associated with left upper quadrant pain.

Painless jaundice, another common presentation, is generally associated with tumors in the pancreatic head or uncinate process. When the tumor does not arise in proximity to the intrapancreatic portion of the bile duct, diagnosis may be delayed and characterized by abdominal pain or back pain without jaundice.

Acute pancreatitis, while uncommon, can be caused by a ductal adenocarcinoma; in patients with no other reason for acute pancreatitis (lack of gallstones, no history of alcohol or precipitating drugs) (²⁸). Symptoms of chronic pancreatitis are relatively common, including diarrhea, bloating or constipation, abdominal distention, and weight loss. Patients with tail lesions often present with signs or symptoms of metastatic disease.

Pancreatic cancer can be difficult to diagnose, particularly in patients with nonspecific complaints. Therefore, patients who present to the oncologist for treatment recommendations may harbor feelings of frustration and anger, having endured a significant delay from the onset of symptoms to diagnosis. Upper endoscopy may have been performed to rule out peptic ulcer disease or other pathology. Endoscopy is seldom helpful unless the pancreatic tumor has invaded the adjacent gastric or duodenal mucosa, leading to ulceration. In this uncommon situation, biopsies may demonstrate adenocarcinoma, and subsequent cross-sectional body imaging reveals an underlying pancreatic mass. Even more rarely, extrinsic compression on the gastric or duodenal wall may be appreciated endoscopically. Unfortunately, upper endoscopy may be misleading and demonstrate mild esophagitis, gastritis, or duodenitis, with or without evidence of Helicobacter pylori. Alternatively, patients complaining of right upper quadrant pain may undergo ultrasonography, potentially revealing gallstones, prompting cholecystectomy. This procedure is usually of temporary benefit and delays the discovery of a pancreatic tumor until pain returns. Last, for patients presenting with complaints of back pain, a musculoskeletal evaluation commonly ensues, with the procurement of plain x-rays, myelograms, or magnetic resonance imaging (MRI) of the spine.

For those patients who present with obstructive jaundice, suspicion of pancreatic cancer is sufficiently high that the diagnostic workup usually proceeds in an orderly fashion with directed imaging studies. These usually include an abdominal ultrasound, computed tomographic (CT) scan of the abdomen, or both. In some centers, discovery of a mass in the head of the pancreas without obvious metastatic disease or evidence for unresectability will prompt an exploratory laparotomy prior to biopsy confirmation of malignancy. This approach is not embraced at MDACC for reasons outlined further in this chapter.

Tissue Acquisition

With rare exception, all patients seen at MDACC are advised to undergo tissue confirmation. Cross-sectional imaging (multidetector CT) should always be performed before interventional endoscopic or radiologic procedures to prevent procedure-related inflammatory changes from confounding assessment. For patients presenting with obstructive jaundice, tissue may be obtained at the time of endoscopic retrograde cholangiopancreatography (ERCP) via brushings of the bile duct at the level of stricture. If brushings are nondiagnostic and CT or MRI suggests that the tumor may be nonmetastatic, we advise endoscopic ultrasound (EUS) with EUS-guided fine-needle aspiration (FNA) (²⁹). This can be performed by experienced operators with minimal risk of duodenal perforation. Moreover, it is thought to decrease the risk of peritoneal or needle-track seeding, which has been reported among patients undergoing transcutaneous ultrasound- or CT-directed biopsies (³⁰). Alternatively, when CT or MRI clearly demonstrates an unresectable, locally advanced cancer, CT- or ultrasound-guided transcutaneous biopsy may substitute for EUS-guided aspiration. If a patient presents with obstructive jaundice and biliary stricture without radiographic evidence of a pancreatic mass, EUS examination is also advised.

When there is radiographic evidence of metastatic disease and an obvious pancreatic mass, we recommend biopsy of a metastatic site, such as the liver. This confirms both the diagnosis and the presence of metastatic disease with one procedure.

Misdiagnosis of Pancreatic Cancer

It is not uncommon for patients to be misdiagnosed with pancreatic cancer. The most common mistake we see is in the setting of bulky peripancreatic adenopathy without a parenchymal pancreatic mass. Adenocarcinomas of the pancreas do metastasize to regional lymph nodes, but lymph nodes are typically small to medium in size. Bulky lymph nodes are seen in other gastrointestinal (GI) malignancies, such as tumors of the esophagus, stomach, duodenum, and occasionally, colon. Lymphoma, non–small cell lung cancer, and carcinomas of unknown primary origin may also lead to bulky peripancreatic lymphadenopathy, thus mimicking a primary pancreatic neoplasm. Thin-cut, dynamic-phase, contrast-enhanced CT will usually rule out the presence of a primary mass in the pancreas in this setting. Another helpful radiographic finding may be the presence or absence of atrophy of the pancreatic body and tail. Although commonly seen in adenocarcinomas originating in the head of the pancreas, this finding is usually absent in the setting of bulky peripancreatic adenopathy, neuroendocrine tumors, and acinar cell tumors. Importantly, patients with neuroendocrine tumors of the pancreas are sometimes misdiagnosed as having a poorly differentiated carcinoma of the pancreas.

High-Quality Computed Tomographic Imaging

The single most important imaging modality is multidetector (multislice) CT. This technique is used to objectively define (anatomically) potentially resectable disease. For optimal pretreatment staging, a CT report in a patient with suspected pancreatic cancer should include the following:

1. The presence or absence of a primary tumor in the pancreas or periampullary region

2. The presence or absence of peritoneal and hepatic metastases

3. Description of the patency of the superior mesenteric vein (SMV) and portal vein (PV) and the relationship of these veins to the tumor

4. Description of the relationship of the tumor to the superior mesenteric artery (SMA), celiac axis, and hepatic artery

Objective radiographic criteria can be used to define a potentially resectable primary tumor of the pancreatic head or uncinate process (Fig. 21-2). The MDACC criteria include: (1) no extrapancreatic disease; (2) a patent SMV and PV (assuming the technical ability to resect and reconstruct this venous confluence); and (3) a definable tissue plane between the tumor and regional arterial structures, including the celiac axis and SMA. Using CT staging and objective criteria for assessment of resectability, many centers have reported resectability rates as high as 75% to 80% (³¹). Of note, CT of the chest is not routinely part of our staging workup. However, if either plain chest x-rays or CT images of the lung bases reveal pulmonary nodules or other suspicious findings, a dedicated CT scan of the chest is obtained. Bone scans and brain imaging are rarely indicated and should not be part of routine staging.

Positron Emission Tomography

Positron emission tomographic (PET) scans are occasionally obtained in the setting of equivocal radiographic findings, such as indeterminate lesions in the liver or lungs. Lesions may be subcentimeter in size, and ¹⁸F-fluorodeoxyglucose negative, even when metastatic disease is present (³²). Scanning by PET is commonly considered when a patient has undergone previous resection and subsequently develops a rising CA19-9 and soft tissue changes in the surgical bed with no other evidence of relapse.

Serum CA19-9 Determinations

CA19-9 measures the specific carbohydrate moiety of the mucin MUC-1 (³³). This is the most commonly elevated tumor marker in pancreatic cancer, but it is not specific and may be elevated in other GI tumors. Whether it should be measured prior to surgery as an independent predictor of resectability or as an adjunct to other clinical staging has not been rigorously studied. Most retrospective analyses generally suggested that a high preoperative CA19-9 level (>500-1,000 IU/mL) implies more advanced disease that is not amenable to resection. We retrospectively analyzed pretreatment CA19-9 levels obtained from 79 patients enrolled in a trial of gemcitabine-based preoperative chemoradiation. All patients had radiographically defined, biopsy-proven, resectable cancer without evidence of metastatic disease. It was found that serum levels greater than 668 IU/mL predicted either the development of overt metastatic disease prior to surgery or early relapse after surgical resection (³⁴). Presently, serum CA19-9 measurements are obtained at presentation to MDACC. When postoperative CA19-9 levels do not normalize within this time frame, it portends early relapse (³⁵). Patients clinically staged as having locally advanced disease but with markedly elevated CA19-9 levels (>5,000) are suspected of having occult metastatic disease. These patients are usually advised to undergo a trial of systemic therapy with serial measurements of CA19-9 levels prior to considering chemoradiation. Improvement of CA19-9 of 50% has been correlated with an improved survival (³⁶).

The Role of Laparoscopy

For patients with potentially resectable disease, some have advocated laparoscopy with biopsies and peritoneal washings as part of routine staging (^37,38). In our experience, fewer than 20% of patients with tumors in the head of the pancreas will have occult metastatic disease when laparoscopy is performed. Given the expense and expected negative findings for 80% of patients, our approach has been to limit this procedure to patients with indeterminate findings on CT (³⁹). An exception applies to the small subset of patients who present with radiographically resectable tumors in the body or tail of the pancreas who are more likely to have occult metastatic disease. The chance of a visible peritoneal metastasis or positive cytology on peritoneal washing is sufficiently high to justify laparoscopy as part of staging (⁴⁰).

The TNM System Versus Clinically Oriented Staging

The TNM (tumor, node, metastasis) staging system for pancreatic cancer is outlined in Table 21-3. For patients undergoing resection, the TNM system is somewhat useful in providing prognostic information and, to a lesser degree, in guiding adjuvant therapy. Generally, patients are staged as having potentially resectable disease, locally advanced unresectable disease, or metastatic disease. For patients with resectable disease who are able to tolerate it, surgery is indicated. Surgery can be preceded by preoperative or adjuvant therapy. Patients with metastatic disease and adequate PS usually receive systemic therapy. For patients presenting with locally advanced disease, treatment should be individualized and may initially involve either chemoradiation or systemic therapy.

Resectable Pancreatic Cancer

It is widely known that surgery holds the only hope of cure for patients with pancreatic cancer. With some exceptions, resectable pancreatic cancers are limited to small tumors in the head of the pancreas. These are removed with a Whipple procedure (⁴¹), more appropriately described as a pancreaticoduodenectomy. Caution is advised when considering a resection of a tail neoplasm; even when these tumors appear localized, they are associated with a higher likelihood of peritoneal seeding compared to a head lesion. Body lesions are almost never amenable to resection.

Unfortunately, there are potential drawbacks associated with up-front surgery:

1. Surgical morbidity and mortality are inversely correlated to experience with the procedure. Several studies have confirmed significant differences in the risk of major perioperative complications and death between hospitals that perform the operation frequently and those that do not (^42,43). Moreover, long-term survival after pancreaticoduodenectomy is improved when performed at a high-volume center (⁴⁴). This is likely attributable to a combination of decreased operative mortality and superior patient selection.

2. Positive surgical margins are associated with a very poor prognosis (see Table 21-4). Surgical margins after pancreaticoduodenectomy can be either microscopically positive (R1 resection) or grossly positive (R2). Median survivals with a positive surgical margin usually range between 6 and 12 months, similar to, if not worse than, the survival of patients with locally advanced disease (^45,46). Many patients undergo laparotomy without adequate preoperative assessment. Some will be found to have unresectable tumors intraoperatively or be left with a grossly positive surgical margin when it might have been possible to predict this prior to surgery. Patients may therefore have a delay in chemoradiation or systemic therapy while the patient recovers. With the exception of patients who present with gastric outlet obstruction or biliary obstruction not amenable to endoscopic or percutaneous stenting, we discourage surgical intervention without high-quality radiographic evidence of resectability. If the surgeon has failed to perform a complete resection, surgery may even be deleterious and compromise the patient’s survival.

3. A substantial proportion of patients do not recover sufficiently to receive postoperative adjuvant therapy. Pancreaticoduodenectomy is a major surgical procedure with removal of portions of the stomach, duodenum, pancreas, and bile duct requiring extensive reconstruction of the upper alimentary canal. Pancreatic anastomotic leaks and delayed gastric emptying are common complications. Retrospective analyses and prospective clinical trials of adjuvant therapy demonstrated that a significant percentage of people do not adequately recover to receive postoperative therapy. For example, Johns Hopkins University demonstrated that of 870 patients who underwent resection for pancreatic cancer with curative intent between 1993 and 2005, only 53% received adjuvant therapy (⁴⁷). Furthermore, analyses of Medicare-eligible patients suggested that among patients 65 years of age or older, fewer than half receive adjuvant therapy (⁴⁸). It is reasonable to assume that a substantial proportion of elderly patients have sufficient difficulty recovering from surgery and this has an impact on adjuvant therapy.

4. Surgically resected patients remain at risk for local failure and metastatic disease. Approximately 80% of resected patients will ultimately relapse and die of disease recurrence. The high risk of relapse stems from an inability to prevent locoregional failure and to eradicate microscopic metastatic disease. Factors predisposing to local recurrence have not been fully elucidated, but recent evidence implicated perineural invasion as an important mediating process. Invasion of nerve sheaths may occur as a pervasive superficial infiltration that cannot be appreciated intraoperatively, even by the most experienced surgeons.

Once patients relapse with distant disease or local failure, no curative strategy is available. Adjuvant therapy, while tending to improve median survival, has not made any significant advances over the past 20 years.

The Role of Adjuvant Therapy

Since the mid-1980s, efforts have been directed toward improving outcomes for patients with resected disease by delivering postoperative adjuvant therapy intended to reduce the risk of relapse and improve long-term survival. Early retrospective analyses of resected patients suggested local failure rates as high as 50% to 80%, which prompted many centers to advocate radiotherapy as a component of adjuvant therapy. The first randomized trial, performed by the Gastrointestinal Tumor Study Group (GITSG), demonstrated a significant survival advantage with chemoradiation based on 5-fluorouracil (5-FU) compared with resection alone (21 vs 11 months) (⁴⁹). The 5-year survival rates were 18% versus 8%, respectively. The European Organization for Research and Treatment of Cancer (EORTC) 40891 trial produced conflicting results 15 years later. In this case, 218 patients receiving 5-FU chemoradiation did not demonstrate a survival advantage over those on observation, although this population was more heterogeneous than that of GITSG because patients with periampullary cancer and those with an R1 resection were included (⁵⁰).

Large randomized adjuvant trials were conducted by the European Study Group for Pancreatic Cancer (ESPAC). In the ESPAC-1 trial, 289 patients were randomized to observation, chemotherapy, chemoradiation, or chemoradiation followed by chemotherapy (⁵¹). Interestingly, chemoradiation was found to have a deleterious effect on survival (median survival 15.9 vs 17.9 months, respectively, P = .05). On the other hand, chemotherapy appeared beneficial over observation, with a median survival of 20.1 months versus 15.5 months (P = .009). As a result of these studies, the role of radiation in adjuvant therapy became controversial. Radiation has been abandoned in the adjuvant setting in many European centers.

After gemcitabine showed superiority to 5-FU in advanced disease, a number of randomized trials tested its role in the adjuvant setting. The Radiation Therapy Oncology Group (RTOG) 9704 trial compared gemcitabine with 5-FU given before and after 5-FU–based chemoradiation. Gemcitabine demonstrated a modest, but not significant, improvement in survival over 5-FU (20.5 months compared to 16.9 months, P = .09) (⁵²).

The benefit of adjuvant gemcitabine was further confirmed with the long-term data from the CONKO 001 trial. Investigators showed a significant improvement in disease-free survival and overall survival with the use of gemcitabine postoperatively (13 and 22.8 months, respectively) compared to surgery alone (6.9 and 20.2 months) (⁵³). ESPAC-3 further compared postoperative gemcitabine to 5-FU (⁵⁴). No statistical difference in survival was noted after a median follow-up of 34 months, although gemcitabine appeared to be better tolerated.

Picozzi and colleagues reported their findings of a phase II study involving adjuvant interferon alpha-2b (IFN-α2b), cisplatin, and continuous infusion 5-FU given concurrently with external beam radiation (⁵⁵). In this study, 89 patients with R0 or R1 resections received IFN-α2b on days 1, 3, and 5 each week for 5½ weeks, cisplatin weekly for 6 weeks, and infusional 5-FU for 38 days with radiation dosed to 50.4 Gy. Overall survival at 18 months was 69% with a median disease-free survival of 14.1 months and overall survival of 25.4 months. Of the patients, 95% experienced grade 3 or greater toxicity. Additional combination trials, however, did not show an improvement in survival compared to 5-FU monotherapy (⁵⁶). Table 21-5 summarizes the adjuvant trials.

Preoperative Therapy for Potentially Resectable Disease

Sadly, there has been no significant progress in adjuvant therapy since the GITSG study was first reported in 1985. More recent studies have been fairly consistent with the GITSG findings: Median survival for resected patients treated with postoperative therapy hovers around 20 months and remains at 12 months for patients undergoing surgery alone. Of the patients who undergo potentially curative surgery, up to 50% do not recuperate enough to begin postoperative chemoradiation or require prolonged recovery to consider treatment. Moreover, rapid disease progression with early systemic relapse is not uncommon after surgery. Therefore, neoadjuvant therapy followed by surgery offers some theoretical advantages over immediate surgery.

Patients who present with potentially resectable disease are generally physiologically fit and make attractive candidates for neoadjuvant therapy. Preoperative therapy allows delivery of chemotherapy or chemoradiation to a relatively well-perfused tumor bed and provides early treatment to microscopic metastases. Positive surgical margins are commonly reported after up-front resection; this is associated with poor prognosis, suggesting that surgery alone provides inadequate local control. Preoperative therapy may provide for sufficient tumor destruction, particularly at the periphery, to increase the chances of a margin-negative resection. Preoperative therapy also allows for observation of the tumor’s underlying biology, and those with aggressive disease are spared a major surgical procedure.

Five preoperative trials have been completed at MDACC (Table 21-6) (^{57,58,59,60,61}). These trials, performed in sequence, have had nearly identical inclusion criteria, with standardized radiographic criteria for resectability, surgical technique, and assessment of resection margins. Our data demonstrated that preoperative therapy is associated with a relatively low local failure rate compared to adjuvant therapy and, over time, modest improvements in overall survival, especially with the use of gemcitabine over 5-FU or paclitaxel-based chemoradiation.

In the work of Evans et al, a total of 86 patients received gemcitabine followed by chemoradiation, and 74% of them were able to undergo pancreaticoduodenectomy (⁵⁷). The median survival was 34 months compared to 7 months for those who did not receive resection. The pattern of failure favored distant metastases; thus, a second trial was designed to increase the amount of systemic therapy. This trial enrolled 90 patients to receive gemcitabine with cisplatin followed by chemoradiation, and 66% underwent resection. The addition of cisplatin did not improve survival beyond gemcitabine alone (31 vs 34 months). While these studies were not designed to be compared to adjuvant trials, the median survival in both preoperative studies was notably better than that seen in the adjuvant data we have to date. In the gemcitabine-based chemoradiation trial, complete pathologic responses were observed in two surgical specimens. While preoperative chemoradiation has not been established as a standard approach, by using preoperative therapy, negative surgical margins are more frequently reported. While these are probably not sufficient to ensure cure, they are likely to be necessary for extended survival (Fig. 21-3).

MDACC Approach to Adjuvant Therapy

At MDACC, adjuvant therapy is delivered with the following principles:

1. Patients must demonstrate adequate recovery from surgery to be considered for further treatment. This includes ample oral caloric intake and no significant impairment of the alimentary tract (delayed gastric emptying, dumping syndrome, uncontrolled pancreatic exocrine insufficiency). Adequate wound healing and absence of infection are also required. Patients should have a PS of 0 to 1.

2. Patients must have adequate hepatic and renal function with sufficient hematologic parameters to undergo cytotoxic therapy.

3. Restaging CT scans are obtained just prior to initiation of adjuvant therapy generally performed 6 to 10 weeks postoperatively. A serum CA19-9 level twice the upper limit of normal precludes patients from enrollment on adjuvant therapy on in-house protocols. Recent retrospective analysis suggested that 5% to 10% of patients who undergo surgery at MDACC will have early radiographic or serologic evidence of relapsing disease prior to initiation of adjuvant therapy. When this occurs, any further therapy is not considered adjuvant.

4. Chemotherapy plus or minus chemoradiation remains the foundation of adjuvant therapy.

At MDACC, patients are encouraged to enroll in postoperative trials of adjuvant therapy. To extrapolate from the experience in locally advanced unresectable disease, patients benefiting from chemoradiation are those who have experienced stable disease with induction chemotherapy. Therefore, our approach at this time includes induction chemotherapy with gemcitabine or a gemcitabine-based doublet for 3 months followed by restaging scans. If no radiographic or serologic evidence of relapse is present at that time, chemoradiation with 5-FU or capecitabine is advised. Radiation is administered in a dose of 50.4 Gy in 28 fractions.

Once postoperative therapy has been completed, patients are followed with restaging CT scans, chest x-ray, physical examination, and standard laboratory tests, including CA19-9 every 6 months for the first 5 years and annually thereafter. A rising CA19-9 after adjuvant therapy does not trigger further systemic therapy until clear evidence of relapse based on physical examination or radiographic studies. Scanning by PET is considered in this situation.

MDACC Approach to Preoperative Therapy

Patients with clinical and radiographic evidence of potentially resectable disease are generally advised to receive protocol-based preoperative therapy, which typically involves chemoradiation. Chemoradiation regimens have varied, and our most encouraging results have been achieved with our gemcitabine-based regimen. After chemoradiation is completed, patients are allowed to recover over 4 to 5 weeks prior to restaging studies. For patients with no clinical or radiographic evidence of metastatic disease and no contraindications to surgery, laparotomy proceeds. At the time of exploration, when no visible evidence of distant disease is encountered, pancreaticoduodenectomy is performed. Postoperatively, further chemotherapy or radiation may be delivered based on the final pathology and the consensus of the multidisciplinary group. Patients are then followed expectantly with periodic restaging studies as outlined previously. Patients who relapse with adequate PS are offered further systemic therapy on or off protocol.

It is important to emphasize that we do not deliver preoperative therapy as a means of staging the primary tumor downward. The medical literature has scattered reports of neoadjuvant therapy being used to successfully stage down patients with locally advanced disease to the point of resectability (⁶²). Caution is advised in interpreting these results because we believe it is possible to stage down patients with borderline or marginally resectable tumors (tumors that abut but do not encase the celiac artery or SMA). These tumors represent a discrete subset; their management, while similar, is more tailored. Figure 21-4 displays an algorithmic approach for resectable pancreatic cancer.

MDACC Approach to Patients With Borderline Resectable Pancreatic Cancer

As high-quality, dynamic-phase, helical CT scanning has developed, an appreciation for the existence of a distinct subset of tumors best described as borderline resectable or marginally resectable has emerged. In this situation, some authorities believe that up-front surgery is more likely to lead to an R1 or R2 rather than an R0 resection. This entity is defined as ≤180 tumor abutment of the SMA or celiac axis, short segment abutment or encasement of the common hepatic artery that is amenable to segmental resection and reconstruction, or short segment occlusion of the SMV, PV, or SMV-PV confluence with a normal SMV below and PV above the tumor to allow for reconstruction (⁶³). Up to 40% of patients with borderline resectable disease have been seen at MDACC, and these patients have a median survival of more than 40 months.

At MDACC, patients with marginally resectable tumors are typically treated with gemcitabine-based chemotherapy for an indefinite period of time, with restaging studies every 2 months. Treatment is continued to maximum benefit, as defined by a nadir in the CA19-9 level or best radiographic response. Thereafter, chemoradiation is delivered, and subsequent restaging studies are performed about 4 to 6 weeks after treatment is complete. Surgery will proceed if there has been some evidence of tumor response, and if no interval development of metastatic disease is apparent, an attempt at surgery will proceed. It remains unclear whether the staging of such tumors downward to technical resectability is of biological significance; therefore, at least 6 months generally elapse at MDACC prior to the contemplation of surgery.

Management of Patients With Locally Advanced Disease

Patients are defined as having locally advanced pancreatic cancer when there is radiographic evidence of SMA or celiac artery encasement, occlusion of the SMV-PV confluence, or significant involvement of the common hepatic artery originating from the celiac trunk. There should be no clinical or radiographic evidence of metastatic disease. Currently, roughly half of all patients present with locally advanced disease. As with resectable pancreatic cancer, an understanding of certain principles will aid in decision making.

1. Locally advanced pancreatic cancer typically progresses over the course of some months. Local tumor progression with worsening pain, new or recurrent biliary obstruction, and gastric outlet obstruction represent difficult management problems. Development of metastatic disease is usually associated with worsening functional status and, unless preceded by a long progression-free interval, is rarely responsive to further therapy.

2. Assessment of response to therapy can be challenging. These tumors may be composed of small nests of adenocarcinoma surrounded by large areas of desmoplasia (Fig. 21-5). Even when cytotoxic therapy is effective, the desmoplastic component of the residual mass may not regress, and the overall tumor mass may appear unchanged. Furthermore, distinguishing the primary tumor mass from surrounding inflammatory changes can complicate the reliable measurement of tumors.

3. All surgical interventions should be considered carefully and be based on PS and life expectancy. Palliative nonsurgical procedures may produce results similar to those of aggressive surgery.

4. One of the primary reasons for considering chemoradiation for patients with locally advanced disease is palliation of pain. However, the clinical benefit associated with chemoradiation has not been rigorously studied. Minsky et al reported significant variations in the estimation of pain relief, with 31% to 77% of patients having improvement in pain after receiving chemoradiation for unresectable disease (⁶⁴).

Data Regarding Chemoradiation Based on 5-Fluoruracil

Support for chemoradiation originates from studies performed by the GITSG. In the original study, patients with locally advanced pancreatic cancer were randomly assigned to receive 40 Gy of radiation plus 5-FU, 60 Gy plus 5-FU, or 60 Gy alone. The median survival was 10 months in each of the chemoradiation groups and 6 months for patients who received 60 Gy without 5-FU (⁶⁵). Of note, these patients had undergone laparotomy and were surgically staged. Only those patients with disease confined to the pancreas and peripancreatic organs, regional lymph nodes, or regional peritoneum were eligible for the study. While this made for a more uniform study population, it also introduced significant selection bias: Enrollment was limited to rapidly recovering patients. In subsequent GITSG studies, neither doxorubicin used as a radiation sensitizer nor multidrug chemotherapy with streptozocin, mitomycin, and 5-FU (SMF) alone or continued after chemoradiation was found to be superior to 5-FU–based chemoradiation (⁶⁶). Additional chemotherapy after 5-FU–based chemoradiation increased toxicity without apparent therapeutic benefit.

In contrast to the results from the GITSG, an ECOG study suggested no benefit of chemoradiation over 5-FU alone (⁶⁷). The ECOG study randomly assigned patients with locally advanced or incompletely resected pancreatic adenocarcinoma to receive chemoradiation (40 Gy and 600 mg/m²/d 5-FU for 3 days) or 5-FU alone (600 mg/m²/week). As in the GITSG studies, all patients were surgically staged and entered in the study within 6 weeks of surgery. The median survival was 8.3 months in the group that received chemoradiation and 8.2 months in the group that received 5-FU alone.

More recent trials of chemoradiation for locally advanced pancreatic cancer have investigated continuous infusion 5-FU in combination with EBRT (external beam radiation). The ECOG performed a phase I study to determine the maximal tolerated dose (MTD) of prolonged infusional 5-FU when combined with EBRT to 59.4 Gy. The MTD of 5-FU was 250 mg/m²/d, with GI toxicity the dose-limiting factor (⁶⁸). A subsequent study conducted in Japan demonstrated the feasibility of utilizing low-dose continuous infusion 5-FU (200 mg/m²/d) over 5.5 weeks combined with a single course of EBRT to 50.4 Gy. This was followed by weekly 5-FU treatments until disease progression. The median survival of treated patients was 10 months, similar to that of patients treated with bolus 5-FU and EBRT in the GITSG trials (⁶⁹). Thus, while infusional 5-FU may provide greater radiosensitivity than bolus 5-FU, no clear survival advantage has been established. In general, for selected patients, treatment programs consisting of EBRT and chemotherapy may result in median survivals of approximately 10 to 12 months and a 2-year survival rate of 20%. Long-term survivors are rare.

Concurrent Chemoradiation Versus Systemic Chemotherapy

Chemoradiotherapy was compared with chemotherapy in a randomized trial by the French Fédération Francophone de Cancérologie Digestive (FFCD) group. In this study, chemoradiotherapy was administered in a dose of 60 Gy concurrently with cisplatin and 5-FU (continuous infusion at 300 mg/m²/d). The chemotherapy arm consisted of gemcitabine (1,000 mg/m²/week). Surprisingly, the overall survival was shorter in the chemoradiotherapy arm (⁷⁰). Higher grade 3 to 4 toxicity rates were observed in the chemoradiotherapy arm compared with the chemotherapy arm (66% vs 40%, respectively), which may partially account for the worse survival.

In 2008, ECOG 4201 compared chemoradiotherapy and chemotherapy alone in a phase III trial. Patients with locally unresectable disease were randomly assigned between chemoradiotherapy with concurrent gemcitabine followed by gemcitabine and gemcitabine alone. In the chemoradiotherapy arm, the total radiotherapy dose was 50.4 Gy with concurrent gemcitabine (600 mg/m²/week). The inclusion of 316 patients was planned, but the study closed after the inclusion of 74 patients because of low accrual. Median overall survival was slightly better in the chemoradiotherapy arm (11 vs 9.2 months, P = .044) (⁷¹). These results should be considered cautiously because of the limited number of patients included. A literature-based meta-analysis concluded that overall survival was not significantly different after chemoradiotherapy or chemotherapy (⁷²).

At the 2013 annual American Society of Clinical Oncology (ASCO) conference, Hammel and colleagues presented the final results of the phase III international LAP 07 study (⁷³). The objective was to determine whether consolidative chemoradiotherapy affected overall survival in patients with inoperable locally advanced pancreatic cancer when tumors were controlled after 4 months of induction gemcitabine-based chemotherapy. Patients were randomly assigned to gemcitabine (1,000 mg/m²/week × 3) or gemcitabine plus erlotinib (100 mg/d) for 4 months. Participants with controlled disease were subsequently randomly assigned to further chemotherapy or chemoradiation (54 Gy [5 × 1.8 Gy/d] and capecitabine 1,600 mg/m²/d). Of the 442 patients initially randomly assigned, 269 patients (61%) entered the second-round randomization phase. A planned interim analysis was conducted after a median follow-up of 36 months and 221 deaths. Median overall survival in the chemotherapy arm was 16.4 months compared with 15.2 months for the chemoradiation group (hazard ratio [HR] 1.03, 95% CI, 0.79-1.34, P = .8295). It appeared neither radiation nor erlotinib improved survival in this population (⁷³). However, administration of radiation did delay the institution of second-line chemotherapy for progressive disease, which had an impact on quality of life.

Integration of Novel Agents Into Concurrent Chemoradiation Strategies

Given the limited benefit noted with 5-FU–based chemoradiation, there has been an effort to incorporate alternative agents into concurrent therapies, including gemcitabine, paclitaxel, capecitabine, and targeted agents, including bevacizumab, cetuximab, and erlotinib. Because of its role in metastatic disease, gemcitabine with EBRT has been extensively investigated for patients with localized cancer. Currently, there is no compelling evidence to suggest improved survival using gemcitabine-based chemoradiation over 5-FU for patients with locally advanced disease. Li et al conducted a small randomized trial that directly compared 5-FU–based chemoradiation with gemcitabine-based chemoradiation. Median survival for the 18 patients randomized to receive gemcitabine with EBRT was 14.5 months, compared with 6.7 months in 16 patients treated with 5-FU. This trial should be interpreted with caution, given the small sample size and poor outcome of patients treated with 5-FU and EBRT (⁷⁴). Another prospective study compared FU with cisplatin-gemcitabine–based chemoradiation and did not demonstrate any difference in overall survival (⁷⁵).

At present, there is no standard approach, dose, or schedule for gemcitabine combined with radiation. Based on completed phase I and II studies, we have defined the MTD of gemcitabine, associated toxicity, and the size of radiation port (⁷⁶). 5-Fluorouracil or capecitabine-based chemoradiation is now standard at MDACC for locally advanced pancreatic cancer.

At MDACC, investigations of novel agents used after chemoradiation have been conducted. In RTOG 0411, patients with locally advanced pancreatic cancer were treated with capecitabine, bevacizumab, and radiation followed by maintenance with capecitabine and bevacizumab. The overall median survival reported was 11 months, which is similar to previous RTOG trials that did not include bevacizumab (⁷⁷).

Systemic chemotherapy alone may improve both pain control and PS and avoids the GI toxicity associated with chemoradiation. For those patients with stable or responding disease after 4 to 6 months of treatment, chemoradiation is often delivered to maximize locoregional tumor control. Chemoradiation is applied only to the patients most likely to benefit as defined by the absence of disease progression during systemic therapy. This strategy was validated by the Groupe Cooperateur Multidisciplinaire en Oncologie (GERCOR) group, who performed a retrospective analysis of patients with locally advanced cancer who received chemoradiation. Investigators noted that 30% of patients developed metastatic disease after induction chemotherapy and were not candidates for radiation. The remaining 70% received continued chemotherapy or consolidative chemoradiation. The overall survival in the two groups was 12 and 15 months (P = .0009) and the progression-free survival was 7 and 11 months, respectively. These data support the strategy of consolidative chemoradiation following induction chemotherapy in patients with locally advanced disease (⁷⁸). Retrospective data from MDACC also strongly suggested that patients who have received induction chemotherapy have a better outcome than those receiving primary chemoradiation (⁷⁹).

MDACC Approach to Locally Advanced Pancreatic Cancer

For patients who have poor PS, supportive care is encouraged, and radiation is contraindicated. In the subgroup of patients with significant pain related to the primary tumor, aggressive use of narcotics is initiated. For patients with poor tolerance of narcotics or inadequate pain control with their administration, celiac or splanchnic nerve block is recommended. Once pain control has improved, therapeutic options are discussed. In our institution, consolidative chemoradiation continues to be used in select cases after an informative discussion with the patient regarding the issues mentioned. When so chosen, at least 3 to 4 months of induction chemotherapy with a gemcitabine-based regimen followed by capecitabine or 5-FU and radiation is the favored approach. Figure 21-6 shows the MDACC protocol for treatment of patients with locally advanced disease.

Management of Metastatic Disease

Compared with patients having other common malignancies, such as cancer of the colon or breast, patients with advanced pancreatic cancer are often much more debilitated. Palliation remains the primary goal of therapy. Management of metastatic disease should be guided by the following principles:

1. The disease course may be quite dynamic, and the clinical status of a patient can change quickly. Patients therefore require frequent reassessment, whether or not they are undergoing cytotoxic therapy.

2. Pancreatic cancer is quite resistant to systemic therapy, and responses to therapy are rarely observed in patients with poor PS or high tumor burden.

3. Peritoneal disease is usually not responsive to chemotherapy and carries a particularly poor prognosis. Metastatic disease predominantly located in the liver or lung is more likely to be responsive to systemic therapy. When the disease is metastatic to the lung only, its course may be somewhat more indolent.

4. Improvement in the treatment for pancreatic cancer is desperately needed, and patients with good PS should be encouraged to participate in clinical trials.

Systemic Therapy for Metastatic Disease—Lessons From the Past

Early published data frequently reported response rates to chemotherapy exceeding 20%. However, with the advent of high-quality CT and MRI, substantially lower response rates have been reported. Importantly, cooperative group studies dating to the 1980s have not clearly demonstrated meaningful survival advantage for patients treated with single-agent chemotherapy compared with 5-FU combinations or even best supportive care. Thus, for many years, no standard drug or drug regimen had emerged as an accepted frontline therapy for metastatic pancreatic cancer.

Gemcitabine for Metastatic Pancreatic Cancer

Chemotherapy for pancreatic cancer changed with the advent of gemcitabine, which was developed in the 1990s. In an early multicenter trial of gemcitabine in 44 patients, 5 objective responses (11%) were documented (⁸⁰). In another study, gemcitabine again led to few objective responses (2 of 32 patients), but symptomatic improvement was also reported (⁸¹). Based on these observations, two subsequent trials of gemcitabine for advanced pancreatic cancer were completed. In the randomized trial that led to gemcitabine’s approval in the United States, weekly gemcitabine was compared to bolus weekly 5-FU in previously untreated patients (⁸²). Patients treated with gemcitabine achieved a higher response rate (5.4% vs 0%) and a statistically significant improvement in median survival compared to those treated with 5-FU (5.65 vs 4.41 months, P = .0025). The 1-year survival rate for gemcitabine-treated patients was 18%, compared to 2% for those treated with 5-FU. Importantly, more clinically meaningful effects on disease-related symptoms were recorded with gemcitabine. This trial enrolled a heterogeneous patient population, with patients having either locally advanced, unresectable disease or metastatic disease. About 70% of the treated patients had metastatic pancreatic cancer, and this is the basis for its use as frontline therapy in patients with disseminated disease.

Fixed-Dose-Rate Gemcitabine

Gemcitabine is a prodrug that is phosphorylated to its active metabolites gemcitabine diphosphate and triphosphate. Gemcitabine diphosphate inhibits ribonucleotide reductase, thereby depleting intracellular pools of the triphosphate nucleotides. Gemcitabine triphosphate can incorporate into an elongating chain of DNA and lead to premature chain termination and cell death. Gemcitabine triphosphate may also inhibit normal DNA repair mechanisms. This may explain its potent radiosensitizing properties and synergy with other DNA-damaging cytotoxic agents.

Once phosphorylated intracellular concentrations are highest when the drug is given at a fixed-dose rate (FDR) of 10 mg/m²/min. A randomized phase II trial in metastatic pancreatic cancer demonstrated that gemcitabine given at 2,300 mg/m² over 30 min compared to 1,500 mg/m² delivered over 150 min (10 mg/m²/min) led to a higher objective response rate (16.2 vs 2.7%) and a trend toward improved survival (6.1 vs 4.7 months) (⁸³). Therefore, when used off protocol, it is administered at an FDR.

Gemcitabine Combinations: Cytotoxic Agents

In an effort to build on gemcitabine for advanced cancer, one approach has been to combine gemcitabine with other cytotoxic drugs. In addition, regimens using two to four other drugs with gemcitabine are reported in the literature. These include combinations of gemcitabine, capecitabine, and docetaxel (GTX) and gemcitabine, 5-FU, leucovorin, irinotecan, and cisplatin (G-FLIP). Randomized trials of gemcitabine versus gemcitabine-based doublets of cytotoxic therapy have shown no statistically significant survival advantage (Table 21-7). However, gemcitabine combined with a platinum does appear to have some benefit in patients with good PS (⁸⁴).

Recent Trials Evaluating Combination Therapy

Cancer and Leukemia Group B (CALGB) 89904 was a four-arm phase II study comparing FDR gemcitabine with the gemcitabine doublets cisplatin, docetaxel, and irinotecan. Six-month survival, the primary end point, did not differ significantly between the four arms (range 53%-57%) (⁸⁵). Overall survival was also similar across groups. A phase III trial combining gemcitabine and cisplatin had a non–statistically significant improvement in progression-free and median survival over single-agent gemcitabine (median survival 7.5 vs 6.0 months, P = .15) (⁸⁶). Similarly, a phase III trial evaluating the combination of gemcitabine and irinotecan versus gemcitabine alone failed to demonstrate a survival advantage over gemcitabine (⁸⁷).

Previously, the GERCOR/GISCAD (Italian Group for the Study of Digestive Tract Cancer) phase III trial with FDR gemcitabine and oxaliplatin demonstrated a statistically significant higher response rate and progression-free survival in patients with locally advanced and metastatic disease; however, overall survival did not reach statistical significance (9.0 vs 7.1 months, P = .13) (⁸⁸). ECOG 6201 enrolled 832 patients in three arms: gemcitabine in a 30-minute infusion, FDR gemcitabine, and FDR gemcitabine with oxaliplatin. Overall survival was not statistically different between the three groups: 4.9, 6.2, and 5.7 months, respectively (⁸⁹).

It was not until September 2013 when the Food and Drug Administration (FDA) first approved a gemcitabine-based combination chemotherapy regimen with nab-paclitaxel. This was based on a follow-up phase III study published in the New England Journal of Medicine that showed an overall survival advantage of gemcitabine plus nab-paclitaxel over gemcitabine alone (8.5 months vs 6.7 months, HR for death, 0.72; 95% CI, 0.62 to 0.83; P < .001). The 1-year survival rates were 35% versus 22%, respectively. The response rate according to independent review was 23% versus 7% in the two groups (P < .001). The combination arm did have increased neutropenia, fatigue, and neuropathy (⁹⁰).

Capecitabine, the orally bioavailable fluorinated pyrimidine, when combined with gemcitabine, demonstrated a modest clinical benefit over gemcitabine alone and appeared to improve median overall survival in patients with good PS (⁹¹). In a phase III trial, Cunningham and colleagues randomized patients to receive gemcitabine versus gemcitabine plus capecitabine (gemcitabine 1,000 mg/m² IV weekly × 3 every 4 weeks; capecitabine 1,660 mg/m²/d by mouth for 3 weeks and 1 week’s rest) (⁹²). The addition of capecitabine to gemcitabine significantly improved overall response rate and progression-free survival (P = .03 and .004, respectively) and trended toward an improved overall survival (P = .08).

Based on preclinical data showing effectiveness of combination chemotherapy in solid tumors, investigators tested a combination chemotherapy regimen consisting of oxaliplatin, irinotecan, fluorouracil, and leucovorin (FOLFIRINOX) as compared with gemcitabine. There was random assignment of 342 patients with an ECOG PS of 0 or 1 to receive FOLFIRINOX (oxaliplatin, 85 mg/m² body surface area; irinotecan, 180 mg/m²; leucovorin, 400 mg/m²; and fluorouracil, 400 mg/m² given as a bolus followed by 2,400 mg/m² given as a 46-hour continuous infusion every 2 weeks) or gemcitabine at a dose of 1,000 mg/m² weekly for 7 of 8 weeks and then weekly for 3 of 4 weeks. The primary end point was overall survival. The median overall survival was 11.1 months in the FOLFIRINOX group as compared to 6.8 months in the gemcitabine group (HR for death, 0.57; 95% CI, 0.45 to 0.73; P < .001). The objective response rate was also improved in the FOLFIRINOX group, 31.6% versus 9.4% (P < .001). More adverse events were noted in the FOLFIRINOX group. At 6 months, 31% versus 66% of patients in the FOLFIRINOX versus gemcitabine group had a definitive degradation of the quality of life (HR, 0.47; 95% CI, 0.30 to 0.70; P < .001) (⁹³).

Molecular Therapeutics for Pancreatic Cancer

While other cytotoxic drugs may provide some survival benefit when combined with gemcitabine, patient outcomes are predicted to be relatively small. Therefore, the investigation of targeted molecular therapies should be given priority. Treatment strategies being developed include interruption or modulation of known growth factors and signal transduction pathways involved with cell growth, invasion, and angiogenesis.

Epidermal Growth Factor Receptor Inhibition

Antibodies to the EGFR have been shown to compete with the growth-stimulatory ligands for binding to this receptor. Small molecular inhibitors of the tyrosine kinase activity of the EGFR have also been developed in a variety of solid tumors, including pancreatic cancer. In a large international phase III trial, erlotinib, an oral small molecule inhibitor of EGFR, in combination with gemcitabine led to a slightly longer median survival compared to gemcitabine alone (6.24 vs 5.91 months, P = .038) (⁹⁴). Importantly, treatment-related toxicities were not significantly worse for the patients receiving the combination compared to monotherapy. This trial resulted in FDA approval for erlotinib in metastatic pancreatic cancer; it remains the only targeted agent approved for this disease to date. Subsequent studies combining the anti-EGFR antibody cetuximab have been less successful. Southwest Oncology Group (SWOG) S0205, a phase III trial evaluating gemcitabine with cetuximab, reported an overall survival of 6 months for gemcitabine versus 6.5 months for the combination (P = .14). Progression-free survival and response rates were similar between the arms (⁹⁵).

Antiangiogenic Agents

Tumor angiogenesis is important in the establishment and progression of metastatic implants. It is now generally accepted that inhibition of these factors represents a feasible approach to impeding metastasis. One cytokine believed to be central to angiogenesis is VEGF, which is often overexpressed in pancreatic cancer. Inhibition of VEGF may have two roles: blocking VEGF receptors to inhibit tumor growth and impeding angiogenesis.

Bevacizumab, an anti-VEGF antibody, has been investigated in patients with advanced pancreatic cancer. Phase II data demonstrated promising response rates ranging from 11% to 24% and overall survivals of 8.1 to 9.8 months (⁹⁶). Unfortunately, a randomized phase III trial, CALGB 80303, did not mirror these results. Patients were assigned to receive gemcitabine alone versus gemcitabine plus bevacizumab (10 mg/kg). Median overall survival was similar (5.2 vs 5.8 months), with no significant improvement in response rate or progression-free survival (⁹⁷). More recently, a multicenter randomized phase III trial with gemcitabine, bevacizumab, and erlotinib was reported and also did not show a significant improvement in survival when compared to gemcitabine, erlotinib, and placebo (7.1 vs 6.0 months; HR 0.89; 95% CI, 0.74-1.07; P = .2087) (⁹⁸). There was, however, a significant improvement in progression-free survival (4.6 vs 3.6 months; HR, 0.73; 95% CI, 0.61-0.86; P = .0002).

Insulinlike Growth Factor Type 1 Targeted Therapies

The IGF-1 clinical trial with ganitumab showed a trend toward improved 6-month and overall survival when combined with gemcitabine as compared with gemcitabine plus placebo (⁹⁹). On the other hand, the addition of cixitumumab to gemcitabine plus erlotinib did not improve survival when compared with gemcitabine and erlotinib (¹⁰⁰). Our randomized phase II study of dalotuzumab showed promising activity when combined with gemcitabine as compared with gemcitabine plus erlotinib. Final analysis of correlative studies from this trial is under way.

Stromal Reengineering and Targeted Therapy

The traditional view of pancreatic cancer stroma has been as a hindrance to delivery of chemotherapy and accounting for the adverse prognosis associated with this cancer. Hedgehog inhibitors were particularly effective in causing stromal depletion (¹⁰¹). This theory, however, was disproven in the clinical setting, with randomized trials of two hedgehog inhibitors, IPI-926 and visomdegib, failing to improve survival when added to gemcitabine as compared with gemcitabine alone. Kalluri et al, from MDACC, recently showed that stromal depletion in genetic engineered mouse models resulted in accelerated tumor growth. In addition, they showed that stromal-depleted pancreatic cancers were sensitive to the CTLA-4 (cytotoxic T lymphocyte-associated antigen 4) antibody ipilimumab (¹⁰²). Similar findings were reported by Rhim et al, who demonstrated that stroma is protective in pancreatic cancer, and deletion of sonic hedgehog accelerated tumor growth; this effect was again reproduced by treatment with smoothened inhibitor (²³). Stroma as a target for therapy continues to be investigated in the clinic.

A recently investigated stromal component is the extracellular matrix component hyaluronan. Enzymatic depletion of hyaluronan by the pegylated hyaluronidase (PEGPH20) resulted in inhibition of cancer growth and prolonged survival when combined with gemcitabine in preclinical studies (¹⁰³). Currently, PEGPH20 is at an advanced stage of clinical development in combination with gemcitabine and nab-paclitaxel.

Second-Line Therapy for Pancreatic Cancer

There is an increasing recognition that second-line therapies have been inadequately researched in pancreatic cancer, representing an important avenue for drug development. Irinotecan liposome injection (MM-398) is a nanoliposomal encapsulation of irinotecan that has been successfully combined with 5-FU in a phase III study (NAPOLI-1) in the second-line setting for pancreatic cancer. The MM-398 with 5-FU and leucovorin achieved an overall survival of 6.1 months compared to 4.2 months’ survival with 5-FU and leucovorin in those who progressed on gemcitabine (HR = 0.67, P = .012) (¹⁰⁴).

Another exciting development in second-line therapy is with ruxolitinib, a JAK1/2 inhibitor that is approved for the treatment of myelofibrosis. In the randomized phase II trial, patients with metastatic pancreatic cancer who had progressed on first-line gemcitabine received capecitabine with ruxolitinib or placebo (¹⁰⁵). On subgroup analysis, patients with a high C-reactive protein level—who represented 50% of the total patient population—had a 6-month survival rate of 42% compared with 11% with placebo (HR = 0.47; P = .005). Interestingly, patients on ruxolitinib also experienced improved weight gain.

Immunotherapy for Pancreatic Cancer

Immune therapy has finally come of age, and immune targeting is rapidly changing the course of multiple cancers. Both innate and adaptive immune systems are involved in the immunosurveillance mechanisms, which include cytotoxic CD8 T cells, T helper 1 (Th1) cells, dendritic cells, tissue macrophages (M1), and natural killer cells. Cancers must escape these surveillance mechanisms to grow and have a clinically significant impact (¹⁰⁶). Preclinical models of pancreatic cancer have informed us that immunosuppressive tumor-associated macrophages (TAMs), Treg cells, along with scarce effector (CD8) T cells occur at even the earliest preinvasive stages and persist through the development of invasive cancer. High concentrations of CD8 T cells, when infrequently present in pancreas cancer, are associated with a good prognosis (¹⁰⁷).

Current immunotherapy approaches for pancreatic cancer have yielded promising results that are being investigated in clinical trials. These approaches include checkpoint inhibitors, pancreatic cancer vaccines, adoptive T-cell transfer, monoclonal antibodies acting at the immune checkpoint level, cytokines, and Treg depletion. Immune checkpoint targeting has the potential of changing the treatment paradigm for melanoma, non–small cell lung cancer, and gastric cancer. These agents either alone or in combination are currently in clinical trials.

In regard to checkpoint inhibitors, ipilimumab has been investigated in 27 cases of pancreatic adenocarcinoma, and one delayed response occurred. The PD1 antibody was studied in an expansion cohort of pancreatic cases (n = 14) without any therapeutic responses (¹⁰⁶). These data, while discouraging, have highlighted the fact that predictive criteria for checkpoint inhibitors are needed. The CD40 agonist was combined with gemcitabine in a study, with four objective responses seen (of the 22 patients treated), and a greater number had metabolic responses on FDG-PET imaging (¹⁰⁸). Anti-OX40 antibodies and IDO inhibitors are currently undergoing testing as well.

Algenpantucel-L, also known as hyperacute pancreatic adenocarcinoma vaccine, is a composite of two irradiated, live, human allogeneic pancreatic cancer cell lines that express murine α-1,3-galactosyl transferase. This results in α-galactosylated epitopes on cell surface proteins, which result in an immune response. A phase II study of this agent, in combination with gemcitabine and radiotherapy, for resected pancreatic cancer resulted in a 1-year overall survival of 86% and disease-free survival of 62% (¹⁰⁹). An increased anti–calreticulin antibody (anti-CALR Ab) level following algenpantucel-L treatment correlated with a statistically significant improvement in overall survival (35.8 months in patients with elevated levels of anti-CALR Ab vs 19.2 months in patients without elevated levels; P = .03). The addition of algenpantucel-L to standard adjuvant therapy for resected patients has been investigated in a phase III clinical study, and the results are anticipated.

Immunotherapy agent GVAX is composed of two irradiated, granulocyte-macrophage colony-stimulating factor (GM-CSF)–secreting allogeneic pancreatic cancer cell lines administered after treatment with low-dose cyclophosphamide (Cy/GVAX) to inhibit Treg cells. The GVAX induces T cells against numerous cancer antigens, including mesothelin-specific T-cell responses. CRS-207 is recombinant live-attenuated Listeria monocytogenes engineered to secrete mesothelin into antigen-presenting cells. These vaccines demonstrated synergistic activity in both antigen-specific T-cell induction and antitumor activity in preclinical models. These two vaccines were investigated in a randomized phase II trial for previously treated patients. This randomized study demonstrated that Cy/GVAX followed by CRS-207 significantly improved overall survival as compared with Cy/GVAX alone in patients with metastatic pancreatic cancer (6.1 months with Cy/GVAX followed by CRS-207 vs 3.9 months with Cy/GVAX alone [HR, 0.59; P = .02]) (¹¹⁰). In this study, mesothelin-specific T-cell immune responses correlated with improved survival. This strategy is now being investigated in a phase III randomized study.

Bruton tyrosine kinase (Btk) is a nonreceptor enzyme of the Tec kinase family that is expressed in B cells, myeloid cells, and mast cells, where it regulates cellular proliferation, differentiation, apoptosis, and cell migration. Inhibition of Btk leads to preferential differentiation of macrophages into M1 instead of immunosuppressive M2 macrophages; Btk inhibition thus decreases the TAMs that promote tumor invasion and metastasis. The BTK inhibitor ibrutinib results in stromal suppression and inhibition of pancreatic tumor growth in preclinical models (¹¹¹). Based on this rationale, the BTK inhibitor, ACP-196, is being investigated in the clinical setting for first- and second-line therapy at MDACC.

Metastatic pancreatic cancer is a disease characterized by anorexia, cachexia, and pain. Therefore, palliation must always be the primary goal for this group of patients and is facilitated by a multidisciplinary approach. Symptomatic relief of biliary obstruction and pain should be addressed prior to consideration of systemic therapy. If pain is not well controlled with oral or transdermal narcotics or if these agents are poorly tolerated, patients should undergo an evaluation with an anesthesiologist or neurologist to consider ablation of the celiac or splanchnic plexus. In addition to aggressive pain control efforts, other supportive measures should be considered, including appetite enhancers, antidepressants, and central nervous system stimulants.

Biliary obstruction should be relieved by nonsurgical means whenever possible, and we advocate the insertion of expandable metal stents rather than polyethylene biliary stents. On occasion, percutaneous biliary drainage may be required in the setting of extrahepatic biliary obstruction.

When a patient develops gastric outlet obstruction, we try to estimate the prognosis at that juncture. If life expectancy is greater than 12 weeks, surgical intervention for definitive gastric bypass is considered. For patients with end-stage metastatic disease, the use of duodenal stents is encouraged. For patients with intractable symptomatic ascites, it is important to realize that this may not be caused by carcinomatosis and frequently results from PV or SMV thrombosis. Ascites secondary to portal hypertension will respond to diuretics, including spironolactone, whereas malignant ascites requires repeated paracentesis or an indwelling peritoneal catheter. Gastroparesis is another commonly occurring problem that requires promotility agents and dietary and behavioral modification.

MDACC Approach to Systemic Therapy for Advanced Pancreatic Cancer

Systemic therapy for metastatic disease should be actively discouraged in patients with poor PS (ECOG >2) or significant metastatic burden. End-of-life discussions are appropriate at the time of diagnosis.

Whenever possible, patients with good PS should be treated with systemic therapy in a clinical trial. The current trial includes the addition of the Btk inhibitor ACP-196 to gemcitabine and nab-paclitaxel for the first-line treatment. In the second-line setting, we are initiating studies with ACP-196, an orally bioavailable, small molecule inhibitor of Btk in combination with the PD1 antibody pembrolizumab.

In terms of nonclinical trial options, after progression on FOLFIRINOX, gemcitabine-based regimens like gemcitabine plus nab-paclitaxel are considered. After gemcitabine and nab-paclitaxel, FOLFOX or single-agent capecitabine are preferred. Patients who have experienced disease stability or response with gemcitabine-based first-line therapy can be considered for second-line therapy with gemcitabine-based combinations (such as gemcitabine, docetaxel, and capecitabine [GTX]) (¹¹²).

For patients who are not candidates for multiagent chemotherapy, gemcitabine as first-line therapy is reasonable. At MDACC, our off-protocol approach is to deliver FDR gemcitabine (600-1,000 mg/m²) at rate of 10 mg/m²/min) weekly. The utility of erlotinib has significantly declined at this time but is a consideration in the first- or second-line setting in combination with gemcitabine. When an objective response or stable disease is observed, chemotherapy is usually continued until there is radiographic or clinical evidence of disease progression, with restaging studies generally performed every 8 to 12 weeks. Gemcitabine-platinum doublets are offered only to those patients with excellent PS and those with BRCA-associated pancreatic cancer.

In summary, clinically meaningful advances in the treatment of metastatic pancreatic cancer have occurred in the past 5 years. These developments have changed the treatment paradigm for patients experiencing modest but significantly improved survival and quality of life. Continued efforts to enroll patients with advanced disease into well-designed clinical trials should remain a high priority for oncologists.