Browse Category: Pancreatic Cancer

Intraoperative radiation therapy


Additional boost irradiation can be given to an unresectable pancreatic cancer during surgery with EB-IORT. The use of EB-IORT with or without EBRT in patients with locally advanced pancreatic cancer has been extensively studied. The use of both EB-IORT and EBRT has been reported to result in improved survival rates compared with surgical bypass alone or EB-IORT alone. This combined therapy may reduce morbidity compared with EB-IORT alone because smaller EB-IORT doses can be used; high single doses of EB-IORT (more than 30 Gy) can result in duodenal ulceration, perforation, or hematochezia from radiation duodenitits. In the United States, several institutions have used EB-IORT at doses between 10 and 20 Gy in combination with preoperative or postoperative EBRT given at doses of 45 to 50 Gy over 5 to 6 weeks. Most studies used EB-IORT at the time of surgical exploration, and in patients with no metastatic disease, treatment was consolidated with postoperative EBRT; median survival times have ranged from 12 to 15 months. Rates of local tumor control appear to be superior to those of historic controls treated with EBRT with or without 5-FU. In a recent report from Mohiuddin and colleagues, 49 patients received EB-IORT (10 to 20 Gy) followed by EBRT (40 to 50 Gy) and systemic chemotherapy (5-FU, 450 mg/m2 ; leucovorin, 100 mg/m2 given every 10 days) given during EBRT and continued for 2 years or until evidence of disease progression. Median survival was 16 months, and the dominant site of recurrent disease was the liver. The extent to which patient selection affected outcome in this and other studies of combined EB-IORT and chemoradiation is difficult to determine.


Patients with clear evidence of encasement of the celiac axis or SMA or occlusion of the SMPV confluence on contrast-enhanced, helical CT do not require laparotomy to confirm that the tumor is unresectable; cytologic confirmation of malignancy can be achieved with CT-guided FNA. This fundamental advance in the pretreatment diagnosis of pancreatic tumors can improve the quality of patient survival and reduce health care costs by avoiding the morbidity and prolonged recovery associated with palliative pancreatic cancer surgery. Published data have demonstrated increased length of survival for patients treated with chemoradiation, but this benefit is limited largely to patients with higher performance status. Therefore, a program of 5-FU-based chemoradiation is justified in fully ambulatory patients with locally advanced disease who have minimal symptoms. Systemic therapy with gemcitabine also represents a reasonable alternative in these patients. For patients with poor performance status, chemoradiation is probably not indicated. Current pharmacologic and interventional techniques for pain control, including percutaneous injection of alcohol into the celiac plexus, have proven highly successful in patients with pancreatic cancer. Furthermore, adequate pain control improves performance status and quality of life, which may translate into increased length of life. The limited therapeutic options available for patients with locally advanced disease and the modest impact of current treatments on survival rates provide the rationale for the entry of patients into trials examining novel systemic agents.


The complex pathophysiologic abnormalities accompanying metastatic pancreatic cancer often make specific treatment decisions extremely difficult. Many patients present to the medical oncologist or surgeon with profound debilitation, severe pain, and extensive metastatic disease. For these patients, intervention with chemotherapy is unlikely to result in significant improvements in quality of life or survival, and the toxic effects of chemotherapy may create additional complications. Management with supportive care or nontoxic hormonal approaches may be the optimal strategies for these patients.

For patients with metastatic pancreatic cancer who present with a good performance status, treatment with systemic chemotherapy is appropriate. In view of the limited impact of the currently available agents on survival, continued enrollment of patients in phase II trials of new agents or combinations is essential. In the absence of access to a phase II trial, treatment with gemcitabine appears to be the evolving standard. However, it must be recognized that the primary impact of gemcitabine is on quality of life; therefore, continued evaluation of novel agents, especially those targeted against specific molecular events important in the pathogenesis of pancreatic cancer, is crucial. As our understanding of the molecular and biochemical basis of pancreatic cancer expands, we will enter a new era in which treatments are tailored to interact with the specific molecular and biochemical targets thought to be important in the development or maintenance of neoplasia.

Treatment of locally advanced disease


A pilot trial of 5-FU and supervoltage radiation therapy in patients with locally advanced adenocarcinoma of the pancreas served as the foundation for a subsequent study of 5-FU-based chemoradiation by the GITSG. All patients were surgically staged; only patients with disease confined to the pancreas and peripancreatic organs, regional lymph nodes, and regional peritoneum were eligible for treatment. The entire area of malignant disease had to be encompassed within a 400-cm2 area. Radiation therapy was delivered as a split course with 20 Gy given over 2 weeks followed by a 2-week rest. Patients received a total of either 40 or 60 Gy. 5-FU was delivered intravenously at a bolus dose of 500 mg/m2 /d for the first 3 days of each 20-Gy cycle and given weekly (500 mg/m2 ) following the completion of chemoradiation. Patients were randomized to receive 40 Gy plus 5-FU, 60 Gy plus 5-FU, or 60 Gy without chemotherapy. Median survival was 10 months in each of the chemoradiation groups and 6 months for patients who received 60 Gy without 5-FU. These data supported the original double-blind study by Moertel and colleagues that compared 35 to 40 Gy of radiation plus 5-FU to radiation alone. Mean survival was 10.4 months in the chemoradiation group and 6.3 months in the group that received radiation alone. Clinically matched, untreated patients with locally advanced pancreatic cancer (retrospectively reviewed from the Mayo Clinic) were also found to have a median survival of approximately 6 months.

All patients were entered in the GITSG studies following laparotomy, at which time the disease was deemed unresectable by the operating surgeon. Chemoradiation was reasonably well tolerated following major surgery. Approximately 80% of patients completed chemoradiation, and the two fatal septic events were believed not to be treatment related. The most frequent toxic effects were nausea and vomiting, which were seldom severe. The significant morbidity reported with palliative pancreatic surgery suggests that only patients with a high performance status could have recovered rapidly enough to be eligible for these studies. Thus, although surgical staging made for a more uniform study population, it also introduced significant selection bias: only rapidly recovering patients were considered for treatment. Comparison of future studies to these data must account for this selection bias. In subsequent GITSG studies, neither doxorubicin (Adriamycin) used as a radiation potentiator nor multidrug chemotherapy (SMF: streptozocin, mitomycin, and 5-FU) alone or continued after chemoradiation was found superior to 5-FU chemoradiation. Additional chemotherapy beyond 5-FU-based chemoradiation increased toxicity without apparent therapeutic benefit.


Treatment planning using high-quality CT allows precise definition of the volume to be treated, enabling the delivery of high-dose EBRT to restricted tumor volumes. The EBRT treatment field encompasses the primary tumor and regional lymph nodes including the celiac axis and the SMA origin. Although these important structures cannot be visualized directly at simulation, their radiographic location is usually at the pedicle of the T12 vertebral body (contrary to the commonly accepted location of the celiac axis at the T12-L1 interspace). In one study, the SMA was found by angiography to arise at the level of the top of L1 in 83% of patients and below the pedicle of L1 in 21%; it arose below the L1-2 interspace in no patients. Because of this anatomic variability, EBRT planning must be individualized by using information from CT, magnetic resonance imaging, or angiographic evaluations. With external irradiation, the isodose lines typically contract in from the superior and inferior field edges, and because of variation in daily treatment set-up, external radiation fields must not encompass T12 or L1 tightly because this risks under-treating the regional lymph nodes.

Treatment simulation is carried out by placing the patient in an arms-up position to avoid an exit dose to the arms. The dose to the primary tumor and regional lymphatics can be specified to the 95% isodose line as a tumor minimum or as an isocentric dose. The limits of normal tissue toxicity guide EBRT to doses of 50.4 Gy (tumor minimum) using fields that rarely exceed 12 × 12 cm. This treatment is usually given in 28 fractions over 5.5 weeks using a four-field technique with anterior-posterior and two lateral fields. Rapid-fractionation irradiation, which delivers 30 Gy (isocenter dose) in 10 fractions over 2 weeks offers the advantages of decreased treatment time, toxicity, and cost. Hyperfractionated irradiation (1.2 Gy twice daily) has also been used for the treatment of unresectable pancreatic cancer, but no improvement in either local tumor control or time to recurrence was found compared with conventional treatment schedules.

Treatment decisions

The use of contrast-enhanced, helical CT allows accurate assessment of local tumor resectability. Objective CT criteria for resectability have replaced exploratory laparotomy as a means of assessing resectability. Pancreaticoduodenectomy should be considered only in patients with a good performance status (Karnofsky 70% or higher) and as part of a multimodality treatment program that includes either preoperative or postoperative chemoradiation. The modest survival rates seen with current treatments (see Table 32.4-2) (Table Not Available) argue strongly for enrollment of all patients into clinical trials evaluating new combinations of surgery, chemoradiation, and newly developed systemic agents. Published perioperative mortality rates support the referral of patients with potentially resectable disease to centers that are experienced with the operative management of pancreatic cancer and that perform at least nine major pancreatic resections per year.
National cancer institute

Current surgical treatment


Current surgical treatment is based on the procedure of pancreaticoduodenectomy as described in 1935 by Whipple and coworkers. Their two-stage pancreaticoduodenectomy consisted of biliary diversion and gastrojejunostomy during a first operation and, after the patient recovered (usually about 3 weeks later), resection of the duodenum and pancreatic head.By 1941, the world experience totaled 41 cases, and the perioperative mortality rate was 30%. Before 1940, the pancreatic remnant was not reanastomosed to the small bowel, and the high mortality rate was largely due to pancreatic fistula from the oversewn pancreatic remnant. In 1941, Whipple modified his reconstruction to include a pancreaticojejunostomy, with the entire procedure done in one operation. In 1946, Waugh and Clagett from the Mayo Clinic described their modification of the one-stage procedure to its current form. The goals of surgical therapy outlined by Waugh and Clagett have not changed in the past 50 years: (1) there should be reasonable opportunity for cure, (2) the risk of death should not outweigh the prospects for cure, and (3) the patient should be left in as normal a condition as possible.

Recent advances in operative technique, anesthesia, and critical care have resulted in a 30-day in-hospital mortality rate of less than 2% for pancreaticoduodenectomy when performed at major referral centers by experienced surgeons. At such centers, mortality rates remain less than 2% despite the use of multimodality therapy, the frequent need for complex vascular resection and reconstruction, and the referral of many patients following an initial unsuccessful attempt at tumor resection. Recently reported mortality rates from other institutions, including university centers and the Department of Veterans Affairs hospitals, range from 7.8% to more than 10%. Data from New York State have demonstrated that hospitals performing fewer than nine pancreatic resections per year have an unacceptably high perioperative mortality rate (12%). Patient outcome will be optimized and costs minimized by the referral of patients requiring major pancreatic resections for malignant disease to centers with active multidisciplinary treatment programs. Surgical resection, however, benefits only patients who undergo a negative-margin resection. Therefore, it is essential that surgery be done only on patients with localized, potentially resectable pancreatic cancer. In the absence of significant innovations in systemic therapy, the only potential for major improvements in the quality of life of patients with pancreatic cancer lies in our ability to limit surgery-related morbidity to those patients most likely to benefit from surgical intervention (i.e., to avoid laparotomy in patients with unresectable disease). Therein lies the importance of how the clinician defines resectability. CT criteria for resectability include the following: (1) the absence of extrapancreatic disease, (2) a patent SMPV confluence, and (3) no direct tumor extension to the celiac axis or SMA.


EB-IORT is a means of delivering a higher dose of radiation to the pancreatic bed and high-risk nodal groups to decrease the risk of local tumor recurrence. The effectiveness of EB-IORT in controlling the primary tumor in patients with unresectable disease, combined with early results from Japan utilizing EB-IORT after pancreaticoduodenectomy, served as the foundation for a study from MDACC of preoperative chemoradiation, extended pancreaticoduodenectomy, and adjuvant EB-IORT. To improve local-regional tumor control in patients with resectable disease, EB-IORT was delivered following resection of the specimen but before initiating gastrointestinal reconstruction. Using a dedicated operating suite containing radiotherapy equipment, the MDACC system of EB-IORT required only an additional 30 to 40 minutes of operating room time because patient relocation was not necessary. The dose of EB-IORT delivered varied from 10 to 15 Gy; these radiation doses were based on preclinical and clinical studies demonstrating the safety of 20 Gy or less. The EB-IORT treatment field included the retroperitoneum and tumor bed extending from the transected bile duct superiorly, to the right kidney laterally, and to the pancreatic remnant medially. The pancreatic remnant and bile duct were excluded from the EB-IORT field. The major retroperitoneal blood vessels (aorta, celiac axis, SMA, SMV, portal vein, and inferior vena cava) included in the EB-IORT field are not as susceptible to radiation injury as are hollow visceral and solid organs. Initial results support the safety of adjuvant EB-IORT and suggest improved rates of local-regional disease control. However, these results are likely multifactorial, being due to accurate preoperative imaging, a standardized approach to surgical resection, and chemoradiation.

European Study Group for Pancreatic Cancer

Additional data regarding the potential benefit of postoperative adjuvant therapy will come from the European Organization for Research and Treatment of Cancer (EORTC) and the European Study Group for Pancreatic Cancer (ESPCA). The EORTC initiated a study in 1987 comparing adjuvant 5-FU-based chemoradiation following pancreatectomy with surgery alone. More than 150 patients have been entered; results are not yet available. In 1994, a study was initiated by the ESPCA randomizing patients following pancreatectomy to one of four treatment groups: (1) no adjuvant therapy; (2) 5-FU-based chemoradiation; (3) 5-FU-based chemoradiation followed by systemic 5-FU and leucovorin; and (4) 5-FU and leucovorin without EBRT.

The risk of delaying adjuvant therapy, combined with small published experiences of successful pancreatic resection following EBRT, prompted many institutions to initiate studies in which chemoradiation was given before pancreaticoduodenectomy for patients with potentially resectable (or locally advanced) adenocarcinoma of the pancreas. The preoperative use of chemoradiation is supported by the following considerations:

Radiation therapy is more effective on well-oxygenated cells that have not been devascularized by surgery.
Peritoneal tumor cell implantation due to the manipulation of surgery may be prevented by preoperative chemoradiation.
The high frequency of positive-margin resections recently reported supports the concern that the retroperitoneal margin of excision, even when negative, may be only a few millimeters; surgery alone may therefore be an inadequate strategy for local tumor control.
Patients with disseminated disease evident on restaging studies after chemoradiation will not be subjected to laparotomy.
Because radiation therapy and chemotherapy will be given first, delayed postoperative recovery will have no effect on the delivery of multimodality therapy, a frequent problem in adjuvant therapy studies.

In patients who receive chemoradiation before surgery, a repeat staging CT scan after chemoradiation reveals liver metastases in 25%. If these patients had undergone pancreaticoduodenectomy at the time of diagnosis, it is probable that the liver metastases would have been subclinical; these patients would therefore have undergone a major surgical procedure only to have liver metastases found soon after surgery. In the MDACC trial, patients who were found to have disease progression at the time of restaging had a median survival of only 6.7 months. The avoidance of a lengthy recovery period and the potential morbidity of pancreaticoduodenectomy in patients with such a short expected survival duration represent distinct advantages of preoperative over postoperative chemoradiation. When delivering multimodality therapy for any disease, it is beneficial, when possible, to deliver the most toxic therapy last, thereby avoiding morbidity in patients who experience rapid disease progression not amenable to currently available therapies.
The survival advantage for the combination of chemoradiation and surgery compared with surgery alone likely results from improved local-regional tumor control. Because of the poor rates of response to 5-FU-based systemic therapy in patients with measurable metastatic disease, it is unlikely that current chemoradiation regimens significantly impact the development of distant metastatic disease. Recent data from Staley and colleagues at MDACC support this belief. Thirty-nine patients received preoperative 5-FU-based chemoradiation, pancreaticoduodenectomy, and electron-beam intraoperative radiation therapy (EB-IORT) for adenocarcinoma of the pancreatic head. Thirty-eight of them were evaluable for patterns of treatment failure; there was one perioperative death. Overall, there were 38 recurrences in 29 patients: 8 (21%) were local-regional (pancreatic bed or peritoneal cavity or both), and 30 (79%) were distant (lung, liver, or bone). The liver was the most frequent site of tumor recurrence, and liver metastases were a component of treatment failure in 53% of patients (69% of all patients who had recurrences). Fourteen patients (37% of all patients; 48% of patients who had recurrences) had liver metastases as their only site of recurrence. Isolated local or peritoneal recurrences were documented in only four patients (11%). This improvement in local-regional control was seen despite the fact that 14 of 38 evaluable patients had undergone laparotomy with tumor manipulation and biopsy before referral for chemoradiation and reoperation. If these 14 patients were excluded, only two patients (8%) would have experienced local or peritoneal recurrence as any component of treatment failure. However, because of the large percentage of patients who developed distant metastatic disease, predominantly in the liver, improved local-regional tumor control translated into only a small improvement in median survival compared with that in other recently published studies. Therefore, in the absence of effective systemic therapy, the goal of chemoradiation (preoperative or postoperative) and pancreatectomy should be to maximize local-regional tumor control while minimizing treatment-related toxicity and cost.

Treatment of potentially resectable disease


Because adenocarcinomas of the pancreatic body and tail do not cause obstruction of the intrapancreatic portion of the common bile duct, early diagnosis is rare; virtually all patients have locally advanced or metastatic disease at the time of diagnosis. CT provides an excellent assessment of the relationship of the tumor to the celiac axis and the SMA origin. Arterial encasement is present in most patients, except for the anecdotal patient who presents with upper gastrointestinal hemorrhage resulting from sinistral hypertension secondary to splenic vein occlusion by a small tumor. In the rare patient who appears to have resectable disease (no arterial encasement and no extrapancreatic disease), laparoscopy before laparotomy is a logical approach because peritoneal metastases are frequently found. The scant data available regarding surgical resection confirm the short survival and poor prognosis in this subgroup of patients. Further, the high 30-day hospital mortality rate (43%; 6 of 14 patients with adenocarcinoma) reported in a study from the Department of Veterans Affairs hospitals suggests that many of the patients taken to surgery have advanced disease and a poor performance status. Accurate preoperative imaging and a selective approach to surgical therapy will minimize treatment-related morbidity and mortality and maximize the length and quality of patient survival.



External-beam radiation therapy (EBRT) and concomitant 5-FU chemotherapy (chemoradiation) were shown to prolong survival in patients with locally advanced adenocarcinoma of the pancreas. Those data were the foundation for a prospective randomized study of adjuvant chemoradiation (500 mg/m2 /d 5-FU for 6 days and 40 Gy of radiation) following pancreaticoduodenectomy conducted by the Gastrointestinal Tumor Study Group (GITSG); that trial also demonstrated a survival advantage from multimodality therapy compared with resection alone. However, because of a prolonged recovery, 5 (24%) of the 21 patients in the adjuvant chemoradiation arm could not begin chemoradiation until more than 10 weeks after pancreaticoduodenectomy.

Further, published studies advocating postoperative adjuvant chemoradiation are prone to selection bias; the patients likely to be considered for protocol entry are those who recover rapidly from surgery and have a good performance status.

The slow patient accrual of postoperative adjuvant therapy studies and the positive correlation of survival with performance status in the GITSG trial validate this concern. A similar selection bias is likely in effect when attempts are made to retrospectively compare patients who received postoperative adjuvant chemoradiation with patients who were treated only with pancreaticoduodenectomy. However, recently reported data from Yeo and colleagues at Johns Hopkins University add further support to the use of multimodality therapy. Those investigators reviewed all patients who underwent pancreaticoduodenectomy for adenocarcinoma of the pancreatic head during a 4-year period. Fifty-six patients received adjuvant chemoradiation, and 22 underwent pancreaticoduodenectomy alone. Despite the chemoradiation group containing a larger percentage of patients with aneuploid tumors, median survival for that group was 20 months compared with 12 months for the group who received surgery alone.

A survival advantage was also demonstrated for patients treated with adjuvant combination chemotherapy alone (5-FU, doxorubicin, mitomycin C) after pancreatectomy. Median survival was 23 months in the 30 patients randomized to receive adjuvant therapy compared with 11 months in the 31 patients treated with surgery alone. Forty-six additional patients were ineligible following surgery, attesting to the difficulty in performing multiinstitution protocol-based research following a complex surgical procedure such as pancreaticoduodenectomy. The toxicity of the surgery and chemotherapy was significant; only 24 of 30 patients received chemotherapy, and only 13 of these received all 6 planned courses of chemotherapy. A previous pilot study of adjuvant 5-FU, doxorubicin, and mitomycin C, using a different schedule of administration, found similar toxicity and therefore questioned the use of adjuvant combination chemotherapy, of even moderate toxicity, after pancreatectomy.

Pretreatment diagnostic studies


If the primary tumor cannot be resected completely, surgery (pancreaticoduodenectomy) for pancreatic cancer offers no survival advantage. However, only 16% to 30% of patients who undergo operation with curative intent have their tumors successfully removed; the remaining patients are found to have unsuspected liver or peritoneal metastases or local tumor extension to the mesenteric vessels. Therefore, most patients who undergo surgical exploration for presumed cancer of the pancreatic head receive no survival benefit, yet the laparotomy results in a perioperative morbidity rate of 20% to 30%, a mean hospital stay of 14 to 20 days, and a median survival after surgery of only 6 months. Further, in patients whose tumors are resected with positive margins (Table 32.4-8) (Table Not Available) , the survival duration is less than 1 year and no different from the survival duration achieved with palliative chemotherapy and irradiation in patients who have locally advanced, unresectable disease. Therefore, in contrast to the case for selected patients with colorectal or gastric cancer, there are no data in support of palliative (positive-margin) resection for adenocarcinoma of the pancreas.


Accurate preoperative assessment of resectability will increase resectability rates and minimize positive-margin resections. A common misconception in pancreatic tumor surgery is that resectability is determined best at laparotomy. In fact, however, resectability is most accurately determined preoperatively by imaging studies, not at the time of “exploratory” laparotomy. Surgeons will declare a tumor to be unresectable at the time of laparotomy when unsuspected liver metastases, peritoneal implants, or, most commonly, locally advanced disease is found. The term locally advanced is often poorly defined, leaving the patient, the medical oncologist, and the radiation oncologist without a clear understanding of why the primary tumor was not resected. Data from MDACC have demonstrated improved rates of resectability when high-quality CT is combined with objective preoperative criteria for resectability. CT criteria for resectability include (1) the absence of extrapancreatic disease, (2) a patent superior mesenteric-portal vein (SMPV) confluence, and (3) no direct tumor extension to the celiac axis or SMA. Patients whose tumors are deemed unresectable by these radiographic criteria are not considered candidates for a potentially curative resection.

Computed Tomography

Improved CT technology over the past decade has resulted in CT being the study of choice to determine extent of disease and resectability status in patients with pancreatic cancer. Image resolution has improved considerably with the use of dynamic scanning, whereby intravenous contrast material is delivered by an automatic injector. The development of helical or spiral scanning has improved scan speed through continuous rotation of the x-ray tube around the gantry. This allows the entire pancreas to be imaged during the bolus phase of contrast enhancement. In addition, scan data can be processed to display images in three-dimensional and multiplanar formats. Helical CT performed with contrast enhancement and a thin-section technique can accurately assess the relationship of the low-density tumor to the celiac axis, SMA, and SMPV confluence. However, design of the scanning protocol and interpretation of scans must be done by experienced physicians who understand the clinical importance of accurate staging and assessment of resectability in patients with pancreatic cancer.

Controversy Over Fine-Needle Aspiration and Intraoperative Biopsy

CT-guided percutaneous fine-needle aspiration (FNA) is the diagnostic procedure of choice for establishing a cytologic diagnosis in patients with locally advanced and metastatic pancreatic cancer. The wisdom of percutaneous biopsy has recently been questioned, and several authors have suggested that the use of preoperative FNA results in peritoneal contamination by tumor cells–contamination that otherwise would not have occurred. However, only in the subgroup of patients with resectable disease are fears of FNA-induced tumor dissemination worthy of consideration; patients who have unresectable disease or who undergo a positive-margin resection have a median survival of only 6 to 12 months, and it is unlikely that microscopic contamination of the peritoneal cavity by tumor cells during biopsy would influence survival of such a short duration. In addition, the majority of patients with unresectable disease require tissue confirmation of adenocarcinoma to enable appropriate counseling and treatment planning. Given the currently available minimally invasive methods of tissue acquisition and biliary decompression, it is unrealistic to consider laparotomy with intraoperative biopsy as an alternative to CT-guided FNA in patients with advanced disease. In patients with potentially resectable disease, the concern about biopsy-induced peritoneal contamination by tumor cells has led to the recommendation that patients with clinical and radiographic findings suggestive of a malignant neoplasm of the pancreas or periampullary region undergo pancreaticoduodenectomy without tissue diagnosis. In one report, CT-guided-FNA was proposed as a cause of positive peritoneal washings in six patients. However, each of these patients appears to have had advanced, unresectable disease, which could have been the source of positive washings. Five of the six reports on the cytologic analysis of peritoneal washings have involved patients with locally advanced or unresectable adenocarcinoma of the pancreas. None of the 38 patients found to have positive washings in these five published series underwent a potentially curative pancreaticoduodenectomy; one patient underwent a palliative resection with grossly positive margins.