Combined-Modality Treatment for Operable Pancreatic Adenocarcinoma
Combined-Modality Treatment for Operable Pancreatic Adenocarcinoma
Although in centers where pancreatectomy is performed frequently, associated morbidity and mortality rates have improved, long-term outcomes in pancreatic adenocarcinoma patients remain poor when surgery is the sole therapeutic modality. The impact of adjuvant chemotherapy on survival in patients with localized pancreatic cancer remains incompletely defined. The European Study Group for Pancreatic Cancer (ESPAC)-1 trial has suggested that overall survival rates are superior when chemotherapy is added to surgery, even when regimens believed to be relatively ineffective in the treatment of advanced disease are used. The role of radiotherapy given with chemotherapy is also unresolved, but chemoradiation continues to receive consideration in the multimodality approach to localized pancreatic cancer. Enhanced collaboration and increased involvement by pancreatic surgeons have helped in the recruitment of pancreatic cancer patients for large-scale randomized clinical trials in Europe and the United States. Many newer chemotherapeutic agents with efficacy in gastrointestinal cancers have yet to be investigated in the adjuvant and neoadjuvant settings.
Pancreatic adenocarcinoma acccounts for approximately 32,000 deaths per year in the United States and more than 210,000 deaths per year worldwide.[1,2] The disease is characterized by early spread to the regional lymph nodes and the liver. The majority of patients who appear to have localized, operable pancreatic cancer have subclinical liver metastases at the time of diagnosis, even when modern staging studies reveal no evidence of extrapancreatic disease. Outcome is dictated primarily by the extent of disease and performance status at presentation. Pretreatment classification of tumors as localized, locally advanced, or metastatic is essential for treatment planning. The median survival durations for these clinical stages of pancreatic cancer are 11 to 18 months, 10 to 12 months, and 5 to 7 months, respectively. This review examines the oncologic issues related to pretreatment staging, surgical management, and pre- and postoperative adjuvant therapy, with specific emphasis on the interpretation of recent clinical trials. Pretreatment Staging Imaging Studies
- Computed Tomography-Multidetector, multiphase helical computed tomography (CT) is the foundation for clinical staging of pancreatic cancer. Although helical CT is widely available, accurate interpretation and reporting of the tumor-related find- ings remain inconsistent. For optimal pretreatment staging and assessment of operability, a CT report for a patient with a suspected pancreatic or periampullary malignancy should include the following information: (1) Commentary on the presence or absence of a primary tumor in the pancreas or periampullary region;
(2) Commentary on 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 relation- ship of these veins and their venous tributaries to the tumor
(4) Description of the relationship of the tumor to the superior mesenteric artery (SMA), celiac axis, hepatic artery, and gastroduodenal artery
(5) Commentary on the presence or absence of aberrant vascular anatomy If a CT study performed early in the diagnostic evaluation is of insufficient quality to assess these issues, a new CT study using a pancreatic tumor scanning protocol should be performed. Specific, objective radiographic criteria should be used to assess operability. These objective criteria include (1) no extrapancreatic disease, (2) a patent SMV and PV, and (3) a definable tissue plane between the tumor and regional arterial structures including the celiac axis and SMA. Using helical CT staging and objective criteria for assessment of resectability, many centers have reported resectability rates of 75% to 80% at surgery.[6,7] Thus, a single imaging study with accurate interpretation and use of objective criteria can facilitate preoperative assessment of resectability in most patients.
- Laparoscopy-The role of laparoscopy in staging remains undefined. Direct inspection of the peritoneal surface and liver with laparoscopy allows for the identification of metastatic disease, the volume of which is below the threshold for detection by helical CT. Laparoscopy detects helical-CT-occult disease in 4% to 15% of patients. Given the relatively low incidence of positive findings at laparoscopy, most centers utilize this technique selectively in patients who have CT and laboratory findings suggestive but not diagnostic of more advanced disease. Such findings include small volume, low-density liver lesions, low-volume ascites, marked elevation in CA 19-9, or a relatively poor performance status suggestive of more advanced disease.
- Positron-Emission Tomography-The role of positron-emission tomography (PET) in pretreatment staging also remains undefined. Following two reports suggesting that results of 18-fluorodeoxyglucose PET altered clinical management in as many as 43% of patients with pancreatic cancer,[9,10] other reports have failed to confirm this degree of clinical benefit, suggesting instead that PET does not significantly change the clinical approach in most patients staged by helical CT.[11,12] As a consequence, the utility of PET in staging patients with no extrapancreatic disease on helical CT and patients with suspicious but nondiagnostic extrapancreatic CT findings remains unclear. The routine use of PET is also limited by its relatively high cost (relative to other diagnostic tests) and its low specificity for distinguishing between neoplastic and inflammatory causes of distal bile duct obstruction.
CA 19-9 is a tumor-associated antigen that is frequently elevated in the serum of patients with pancreatic adenocarcinoma.[ 13,14] CA 19-9 may also be mildly elevated in certain benign conditions such as pancreatitis, and thus modest serum CA 19-9 elevations cannot be used for the diagnosis of or screening for pancreatic malignancy. In addition, patients lacking the Lewis antigen glycosyltransferase are unable to produce CA 19-9; this deficiency is present in 7% to 10% of the general population.[13,15] Furthermore, the production of CA 19-9 by normal biliary ductal cells is increased in the settings of biliary obstruction and inflammation; serum CA 19-9 levels are influenced by the presence of jaundice. Since the initial serum CA 19-9 level is obtained soon after presentation with clinical jaundice in many patients, CA 19-9 measurements should be repeated after the biliary obstruction and jaundice have been relieved. There is no consensus on what CA 19-9 level is sufficiently reliable for the diagnosis of cancer in patients in whom there is a clinical suspicion of pancreatic adenocarcinoma. In an analysis by Forsmark and colleagues, serum CA 19-9 levels above 90 U/mL and above 200 U/mL had estimated accuracy rates of 85% and 95%, respectively, in the diagnosis of malignancy in the clinical setting of a suspected pancreatic neoplasm. In one study, the combination of CT and CA 19-9 measurement in nonicteric patients yielded a positive predictive value of 99% to 100% in the diagnosis of pancreatic cancer when a CA 19-9 reference value of 100 to 120 U/mL was used, and a report by Tian et al. suggested that an elevation in CA 19-9 above 750 U/mL was associated with a high probability of locally advanced or metastatic disease. As with most tumor markers, CA 19-9 may be optimally used as supporting clinical evidence in the diagnosis and staging of pancreatic cancer. Pretreatment Biopsy The role of pretreatment biopsy in patients who present with localized disease and in whom surgery is anticipated depends on specific patient and physician factors. One school of thought maintains that for patients presenting with a clinical and radiographic picture consistent with a resectable pancreatic neoplasm, the surgical resection should be both diagnostic and therapeutic. This approach assumes that positive intraoperative biopsies are not required before proceeding with pancreatic resection and that diagnostic pancreatectomy can be performed with low morbidity and mortality rates. A second school of thought espouses reasonable efforts to achieve a pretreatment tissue diagnosis. This approach seeks to separate the diagnostic and therapeutic phases of pancreatic cancer management because doing so may have specific advantages for patients and physicians. A pretreatment biopsy diagnosis allows the patient to consider referral to a regional center specializing in pancreatic cancer treatment. In contrast, when surgery for both diagnosis and treatment is offered, the patient is usu- ally not able to investigate options for further staging or for referral until after surgical staging with or without tumor resection. At this point, an important therapeutic bridge has been crossed. As outlined below, consideration of referral to a regional center is well founded given the significant differences in operative mortality and longterm survival observed between higher volume regional centers and smaller hospitals. Fine-needle aspiration guided by CT or endoscopic ultrasonography is a relatively safe and accurate means of pretreatment tissue diagnosis. Pancreatectomy Technique
Subtotal pancreatic resection can consist of a left (distal) pancreatectomy or right pancreatectomy, which is usually anatomically described as a pancreaticoduodenectomy or Whipple procedure. For most malignant tumors arising in the pancreatic head and uncinate process, pancreaticoduodenectomy is required to achieve microscopically negative pancreatic parenchymal and retroperitoneal surgical margins. Standard pancreaticoduodenectomy involves resection of the distal stomach, distal bile duct (with cholecystectomy), duodenum, pancreatic head and uncinate process, proximal jejunum, and regional lymph nodes. Gastrointestinal reconstruction following pancreaticoduodenectomy requires enteric, biliary, and pancreatic anastomoses, usually by gastrojejunostomy (or duodenojejunostomy in the case of pylorus-preserving pancreaticoduodenectomy), choledochojejunostomy, and pancreaticojejunostomy (or pancreaticogastrostomy). Technical modifications that have no clearly demonstrated advantage or adverse oncologic impact include pylorus preservation and extended lymphadenectomy.[24,25] Randomized trials of pancreaticoduodenectomy with extended vs standard lymphadenectomy have demonstrated increased morbidity with the extended dissection but no improvement in survival rates.[24,25] As a consequence, most Western surgeons perform either a standard or pyloruspreserving pancreaticoduodenectomy with a standard lymphadenectomy (including peripancreatic lymph nodes) but not extended regional lymphadenectomy. The most oncologically significant and technically demanding step in pancreaticoduodenectomy is the retroperitoneal vascular dissection along the proximal SMA. This step occurs during the final phase of the tumor resection, after division of the stomach, jejunum, bile duct, and pancreas. This step can be technically difficult and associated with a risk of hemorrhage from the pancreaticoduodenal arteries arising from the right posterolateral side of the SMA. In an effort to minimize this risk and save time, many surgeons do not fully mobilize the SMV to adequately expose the SMA and instead choose an easier dissection plane several millimeters to the right of the SMA. Unfortunately, this compromises the radial retroperitoneal margin, which is of critical oncologic significance. Pancreaticoduodenectomy with segmental resection of the SMV or PV is necessary when the tumor is inseparable from the lateral wall of the SMV or PV. When intraoperative findings suggest venous involvement, approximately 80% of patients undergoing pancreatectomy with venous resection have histologic evidence of tumor extension into the vein wall. However, venous resection should be performed only in carefully selected patients who have tumor adherence to the SMV or PV but have no evidence of tumor extension to the SMA or celiac axis (Figure 1). Following pancreaticoduodenectomy with venous resection, patients with pathologic invasion of the SMV or PV do not have an increased frequency of surgical margin or lymph node positivity compared to patients without involvement of the SMV or PV, and patient survival after pancreaticoduodenectomy with venous resection is comparable to that observed in patients with similar tumors that do not require venous resection. Hospital Pancreatectomy Volume and Mortality Rates
Several reports from Europe[28,29] and North America[30-32] have established a clear relationship between institutional pancreatectomy volume and perioperative mortality rates. For North American patients, the recent analyses of national databases by Birkmeyer et al and Kotwall et al of 10,507 and 25,000 patients who underwent pancreatectomy are among the most compelling of these reports. Birkmeyer and colleagues demonstrated that patients undergoing pancreatectomy at institutions with very low (< 1 pancreatectomy per year), low (1 or 2 per year), or moderate (3-5 per year) volumes had mortality rates of 16.2%, 14.4%, and 10.9%, respectively. In comparison, patients undergoing pancreatectomy at higher-volume (> 16 per year) institutions had a mortality rate of only 3.9%. Kotwall et al reported similar findings in an analysis of surgical volume and postoperative mortality rates at hospitals categorized as rural, urban nonteaching, and urban teaching. In that report, the overall national operative mortality rate for pancreaticoduodenectomy was 14%, and there was a clear relationship between perioperative and postoperative mortality rates and hospital surgical volume and category. Data from these two reports demonstrate that the majority of patients undergoing pancreatic resection in the United States have surgery in hospitals where the pancreatectomy volume is low or moderate and the mortality rate is in excess of 10%. Higher-volume hospitals also have improved long-term cancer-related outcomes. Birkmeyer et al performed a retrospective cohort study of all 7,229 patients over age 65 years who underwent pancreaticoduodenectomy in the United States between 1992 and 1995. The 3-year overall survival rates were higher at high-volume centers (37%) than at moderate- (29%), low- (26%), and very lowvolume hospitals (25%, P < .001). After exclusion of perioperative deaths and adjustment for case mix, patients who underwent surgery at high-volume centers remained less likely to experience late deaths than did patients at very low-volume hospitals (adjusted hazard ratio [HR] = 0.69; 95% confidence interval [CI] = 0.62- 0.76). The reasons for this apparent improvement in long-term outcome are not evaluable from this type of analysis of administrative databases but are presumed to be related to improved staging and possibly increased use of adjuvant treatments in these patients. The data establishing a definite relationship between institutional volume and short- and long-term survival rates suggest that a public health policy promoting regionalization of care for patients with operable pancreatic cancer would immediately improve survival rates.[34,35] Indeed, the survival advantages associated with regionalization of care may approach or exceed those available with current adjuvant treatments. Complications and Postpancreatectomy Recovery
Although mortality rates for pancreaticoduodenectomy have declined to less than 3% in high-volume centers,[36-38] the morbidity of pancreatectomy remains high, with pancreatic leak and infection among the most common surgical complications.[ 39] Postpancreatectomy recovery can be prolonged owing to the complex interplay between comorbid factors and surgical morbidity. Unfortunately, prolonged recovery following pancreatectomy may delay and in some cases prevent planned adjuvant treatment. Reports from major referral centers for pancreatic surgery demonstrate that approximately 25% of postpancreatectomy patients do not recover satisfactorily (ie, achieve a good enough performance status) to consider postoperative chemoradiation.[ 40,41] However, this figure should be considered a minimum estimate of the impact of delayed postoperative recovery, as the majority of patients in North America undergo pancreatectomy at low- to intermediate- volume institutions, where morbidity (as inferred from mortality data) and mortality rates are considerably higher than at regional centers.[30,31] Many medical and radiation oncologists have little familiarity with normal postpancreatectomy recovery and the specific intermediate-term complications that can affect a patient's performance status at the time of assessment for adjuvant treatment. For patients under age 70 with minimal comorbidity, at least 6 to 8 weeks is generally required for recovery to a performance and nutritional status that is satisfactory for consideration of postoperative treatment. Clinically significant issues that are frequently present in variable degrees of severity at the time of consultation for adjuvant treatment include low-grade delayed gastric emptying, as manifested by occasional nausea or vomiting, and pancreatic exocrine insufficiency, manifested by frequent bowel movements with bulky or greasy stools and difficulty maintaining or gaining weight. A trial of metoclopramide is often considered for patients with clinical evidence of delayed gastric emptying. In addition, pancreatic enzyme supplements should be used in almost all patients. These enzyme supplements should be titrated to reduce bowel movement frequency to one or two per day. H2-receptor blockers or proton pump inhibitors should also be used in all patients to minimize the risk for marginal ulceration at the gastrojejunal anastomosis, prevent treatment- related gastritis if radiotherapy is planned, and facilitate the action of pancreatic exocrine supplements, which require an alkaline environment for activation. Classification and Oncologic Significance of Surgical Margins
For tumors arising in the pancreatic head and uncinate process, the retroperitoneal margin is defined anatomically as the margin between the medial edge of the tumor and the right lateral border of the SMA (Figure 1). During pancreaticoduodenectomy, dissection must occur in the immediate periadventitial plane of the SMA to maximize the radial retroperitoneal margin. As a consequence, the SMA dissection is the most oncologically significant step in pancreaticoduodenectomy. Difficulties in maximizing and pathologically evaluating the retroperitoneal margin include the technical challenge of safe dissection along the proximal SMA and the fact that identification and orientation of the specimen for accurate margin assessment is usually possible only at the time of surgery by combined efforts of the surgeon and surgical pathologist. This margin should be inked in the surgical suite, and the pathology report should describe the distance in millimeters between the tumor and the inked retroperitoneal margin; retrospective evaluation of this margin is not possible. Details of the procedure for margin assessment are outlined in the sixth edition of the AJCC Cancer Staging Manual. Pancreatectomy margins (especially the retroperitoneal margin) should be classified by the surgeon after integration of the operative findings and the microscopic surgical margin findings in the final pathology report. All pancreatic resections should be class- ified according to residual disease status (termed "R" factor): R0, no gross or microscopic residual disease; R1, microscopic residual disease (microscopically positive surgical margins with no gross residual disease); and R2, grossly evident residual disease. The pathologist cannot usually differentiate an R1 (microscopically positive) from an R2 (grossly positive) retroperitoneal margin in the absence of information regarding the retroperitoneal dissection, which should be included in the operative note. The R designation should appear in the final pathology report and should be consistent with the dictated operative note. For example, if the surgeon states that gross tumor was encountered when completing the retroperitoneal dissection, a positive histologic margin should result in the R2 designation in the final pathology report. In the absence of this information being included in the operative report, the proper R designation cannot be determined. The difficulty in differentiating R1 from R2 resections has significant implications for clinical trials examining the potential survival advantage of adjuvant or neoadjuvant therapies. The limited data available suggest that the radial retroperitoneal margin is positive in 15% to 25% of patients after pancreaticoduodenectomy (single- institution data). The clinical significance of a positive margin is indicated in Table 1. Surgical resection with a grossly positive surgical margin is associated with a median survival of 6 to 12 months. This median survival in patients with ostensibly localized pancreatic cancer is comparable to that observed in patients with locally advanced pancreatic cancer treated nonsurgically.[ 44,45] Thus, a positive-margin (R2) pancreatic resection represents a suboptimal therapeutic result. Because of the lack of accurate historical recording of R status, the impact of an R1 resection-especially in the era of contemporary adjuvant therapy-is unknown. Oncologic Outcome After Pancreatectomy Alone
Unfortunately, despite improvements in the short-term outcomes of patients undergoing pancreatic resection for pancreatic cancer, there has been no significant improvement in oncologic outcome over the past decade when pancreatectomy is used as the sole therapeutic modality. Many surgical series have relatively short median follow-ups and/or do not report sufficient information on the use of nonsurgical treatment to draw firm, accurate conclusions regarding outcome in patients treated by pancreatectomy alone. However, retrospective reports of surgical outcome stratified by additional treatments and results from the pancreatectomy-alone arm of prospective randomized trials of pancreatectomy alone vs pancreatectomy plus postoperative chemoradiation suggest that the median survival following pancreatectomy alone (without chemoradiation) for localized pancreatic adenocarcinoma ranges from 11 to 17 months (Table 2).[46-49] That said, the use of contemporary pretreatment imaging and improved patient selection may increase the median survival a few months over these historical estimates. For purposes of treatment comparisons and protocol planning, median survival durations of 12 to 18 months and 5-year actuarial overall survival rates of 10% to 15% with pancreatectomy alone appear to be accurate estimates. On this basis, it is clear that for most patients, pancreatectomy is necessary but not sufficient for cure of pancreatic cancer.