|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Phase I Trial of Radiation Dose Escalation With Concurrent Weekly Full-Dose Gemcitabine in Patients With Advanced Pancreatic CancerByFrom the Department of Radiation Oncology, University of Michigan, Ann Arbor; and William Beaumont Hospital, Royal Oak, MI. Address reprint requests to Cornelius J. McGinn, MD, Department of Radiation Oncology, University of Michigan, 1500 E Medical Center Dr, Ann Arbor, MI 48109-0010; email: mcginn{at}umich.edu
PURPOSE: The primary objective of this phase I trial was to determine the maximum-tolerated dose of radiation that could be delivered to the primary tumor concurrent with full-dose gemcitabine in patients with advanced pancreatic cancer. PATIENTS AND METHODS: Thirty seven patients with unresectable (n = 34) or incompletely resected pancreatic cancer (n = 3) were treated. Gemcitabine was administered as a 30-minute intravenous infusion at a dose of 1,000 mg/m2 on days 1, 8, and 15 of a 28-day cycle. Radiation therapy was initiated on day 1 and directed at the primary tumor alone, without prophylactic nodal coverage. The starting radiation dose was 24 Gy in 1.6-Gy fractions. Escalation was achieved by increasing the fraction size in increments of 0.2 Gy, keeping the duration of radiation constant at 3 weeks. A second cycle of gemcitabine alone was intended after a 1-week rest. RESULTS: Two of six assessable patients experienced dose-limiting toxicity at the final planned dose level of the trial (42 Gy in 2.8-Gy fractions), one with grade 4 vomiting and one with gastric/duodenal ulceration. Two additional patients at this dose level experienced late gastrointestinal toxicity that required surgical management. CONCLUSION: The final dose investigated (42 Gy) is not recommended for further study considering the occurrence of both acute and late toxicity. However, a phase II trial of this novel gemcitabine-based chemoradiotherapy approach, at a radiation dose of 36 Gy in 2.4-Gy fractions, is recommended on the basis of tolerance, patterns of failure, and survival data.
CURRENT THERAPEUTIC approaches for patients with unresectable pancreatic cancer include the use of chemoradiotherapy or chemotherapy alone. Unfortunately, the benefits associated with these therapies are modest. Combined-modality treatment most commonly consists of radiation and fluorouracil (5-FU), based on a single study that demonstrated improved median survival with the combination as compared with radiation therapy alone.1 Efforts to improve survival in patients with unresectable disease have often focused on measures to improve local control. These measures include radiation dose escalation using intraoperative electron beam therapy, which has been reported to reduce local failure.2,3 Yet the Radiation Therapy Oncology Group trial found no survival advantage in a multi-institutional setting.4 The delivery of 5-FU as a protracted venous infusion (PVI) has been investigated more recently, in an attempt to optimize radiosensitization.5,6 The impact of this change in the schedule of 5-FU administration is not apparent, and as yet there are no data to indicate improved survival. The failure of these measures to improve outcome has been attributed, in large part, to the development of distant metastases. It is apparent that more effective treatment for pancreatic cancer must simultaneously address both local and distant sites of failure. Toward this end, we designed a gemcitabine-based chemoradiotherapy protocol that differed from other gemcitabine-based combined-modality regimens in two important ways. First, we chose to use standard doses of gemcitabine, considering the clinical benefit associated with its use as a systemic agent.7 The use of a standard doses is consistent with our laboratory data that demonstrate maximum radiosensitization when cytotoxic concentrations of drug are used.8 However, use of full-dose gemcitabine required reduction and investigation of the radiation dose based on prior clinical experience. This differs from the more traditional approach of chemotherapy dose escalation with a fixed dose of radiation. Second, we elected to radiate the primary tumor alone, without the inclusion of normal-appearing regional lymph nodes. This was based on the assumption that the majority of the benefit from radiation would result from control of the primary tumor, rather than control of subclinical disease in these nodes. Regional nodes could potentially be controlled by standard doses of gemcitabine, as would more distant sites. Concerns for excess normal tissue toxicity that might occur with the use of more conventional treatment volumes contributed to this decision. This strategy required more accurate identification of the primary tumor and three-dimensional conformal radiation treatment. In this phase I trial, gemcitabine was delivered at the current recommended dose (1,000 mg/m2 weekly times three every 28 days), with radiation dose escalation in successive patient cohorts. The primary objective was to determine the maximum dose of radiation that could be delivered via conformal therapy directed at the primary tumor during the first of two cycles of gemcitabine. The overall hypothesis is that this type of chemoradiotherapy regimen, which emphasizes systemic therapy, would improve tumor control at all sites and, subsequently, survival. However, it is possible that excess local and regional failures could result from a reduction of the radiation dose and a smaller treatment volume, respectively. Therefore, a secondary objective was to document patterns of failure. Data describing tolerance and indicating both local and regional control would support investigation of this type of chemoradiotherapy regimen further, particularly as more effective systemic therapies are developed.
Eligibility Patients with pancreatic or ampullary adenocarcinoma, unresectable or resected with involvement of surgical margins or regional nodes, were eligible for this trial. Histologic or cytologic confirmation was required. Radiation dose escalation proceeded separately in two groups of patients: those with duodenum in the planned radiation field and those in whom duodenum would not be in the planned radiation field (primarily those treated after pancreaticoduodenectomy). This design was pursued because duodenal toxicity was projected to be the predominant dose-limiting toxicity (DLT) with this combined-modality regimen. This report is based on the cohort of patients with duodenum in the radiation field for whom the radiation dose was escalated, as described below, independently from the cohort of patients in whom duodenum would not be in the planned radiation field. Patients with duodenum in the radiation field included those with unresectable disease and those who had undergone resection of lesions in the body or tail of the pancreas, without resection of the duodenum.
Determination of unresectability was based on helical computed tomography (CT) scan, endoscopic ultrasound, and/or exploratory laparotomy using the criteria defined in the National Comprehensive Cancer Network guidelines for pancreatic cancer.9 Patients with evidence of distant metastases were eligible. Pretreatment evaluation included a complete history and physical examination, baseline assessment of organ function, chest x-ray, and CT scan of the abdomen. Further eligibility included age
Treatment
equation
The occurrence of gastrointestinal DLT (described below) prompted discontinuation of radiation therapy and gemcitabine (if noted on the day of planned infusion). Occurrence of toxicities Three-dimensional radiation treatment planning was used in all cases. Patients were immobilized in a foam cradle in a supine position, and the treatment planning CT was obtained on a helical scanner with both oral and intravenous contrast. The gross tumor volume was the primary tumor identifiable on CT scan. The clinical target volume was defined as the gross tumor volume plus 0.5 cm. The planning target volume was the clinical target volume plus 0.5 cm for daily patient set-up variation. No prophylactic nodal irradiation was given. Treatment planning was performed with the isocenter calculated at 100% and the 95% line encompassing the planning target volume. The spinal cord was limited to a generally accepted tolerance dose, considering the change in fraction size. If one kidney was to receive more than 20 Gy, then more than 90% of the remaining kidney was excluded from the primary beam. Generally, a three-field nonaxial beam arrangement (opposed laterals with an anterior-inferior oblique) was used, as previously described.10
Radiation Dose Escalation
Acute toxicity was defined using the National Cancer Institute common toxicity criteria (version 1). In the initial phase of the trial (dose levels of 24 and 27 Gy), DLT was defined as grade Patients were seen and examined by one of the investigating physicians at least once weekly during the course of radiation therapy and on a minimum of two occasions during the second cycle of gemcitabine. Patients returned for an evaluation of acute toxicity 2 weeks after the last dose of gemcitabine (week 9). This evaluation included a physical examination, weight, laboratory evaluation, and notation of performance status. After two cycles of therapy, as described above, additional chemotherapy could be delivered at the discretion of the treating physician.
Response Assessment and Patient Follow-Up
Statistical Considerations
Patient Characteristics Between September 1997 and May 2000, 37 patients were entered onto this study. The median age was 59 years (range, 42 to 80 years). Three patients were treated after surgical resection of primary tumors in the body or tail; two had gross residual disease, and a third patient had both positive margins and positive nodes. The remaining patients had unresectable disease in the ampulla (n = 1), head (n = 30), or body (n = 3) of the pancreas. Biopsy-proven metastases were noted in six patients at the time of enrollment and suspected in another eight based on radiographic findings. Three patients had received one cycle of gemcitabine alone before study entry. Seventeen patients continued to receive gemcitabine alone after completion of protocol therapy while an additional five patients received gemcitabine and cisplatin. Postprotocol chemotherapy was most commonly delivered for 4 months.
Toxicity Nonhematologic toxicity meeting the criteria for DLT occurred in three patients, including two with grade 4 vomiting in the third week of radiation therapy and one with the development of abdominal pain during the second cycle of chemotherapy, secondary to gastric and duodenal ulceration (Table 1). One DLT occurred at the 30-Gy dose level, whereas the other two occurred at the 42-Gy dose level. Thus at the final planned dose level of the trial (42 Gy), DLT was noted in two of six assessable patients.
Treatment was completed in 35 of 37 patients. One patient progressed while on therapy, and another was removed from study after having experienced a myocardial infarction after the course of radiation therapy but before completion of the second cycle of gemcitabine. This was considered an event unrelated to treatment, because the patient was without evidence of other toxicity at the time. Seventy percent of protocol patients received at least 85% of the gemcitabine dose intended. The median percentage dose received was 96%. There was no apparent relationship between radiation dose level and the percentage gemcitabine dose received, and no difference in the percentage dose received in the first cycle of chemotherapy compared with the second cycle (P > .25). Gastrointestinal toxicity was also assessed by documenting patients’ weight in the final week of radiation therapy (week 3) and again after completion of all protocol therapy (week 9). The mean weight loss, compared with the patients weight immediately before initiation of therapy, was 1.1% (± 0.5%) at week 3 and 1.0% (± 1.1%) at week 9. Weight loss did not significantly differ between radiation dose levels (P = .62). Among the 35 patients who completed the course of protocol therapy, 25 (71%) remained within 5% of their pretreatment weight. Finally, weight gain was noted in 15 patients (44%), excluding one patient who developed ascites. Five patients have experienced events attributable to late toxicity. Two events occurred in patients at the lower range of radiation dose: the development of a bleeding ulcer 3 months after treatment with 24 Gy and the development of renal artery stenosis 18 months after treatment with 27 Gy. The other three events occurred after delivery of 42 Gy: recurrence of gastric/duodenal ulceration that was initially noted during the second cycle of chemotherapy (and scored as a DLT), stricture of the transverse colon 2 months after completion of radiotherapy, and the development of a bleeding duodenal ulcer 7 months after completion of radiotherapy. The latter two patients required surgical management.
Response
Among the 10 patients with objective response, eight were without evidence of distant metastases. Whipple resection was attempted in three of these patients. Two were found to be locally unresectable at the time of exploratory laparotomy. The remaining patient underwent resection with both negative nodes and negative margins. Whipple resection was also attempted in two patients with stable disease, one of whom underwent resection but was found to have lymph node metastases. There has been no evidence of local progression among the patients with an objective response of the pancreatic tumor. Four remain without evidence of any progression 8, 12, 19, and 22 months after initiation of protocol therapy, including the patient who underwent resection after response (at 22 months) and a patient with suspected metastases at presentation (at 12 months).
Patterns of Failure In 14 patients with suspected or proven metastatic disease at the time of enrollment, radiographic or clinical evidence of failure at any site has been noted in 12. In this group, one patient has experienced local failure, two patients have developed regional adenopathy, and all 12 have experienced disease progression at distant sites (liver, peritoneum, and lung).
Survival
The primary objective of this trial was to determine the MTD of radiation therapy that could be delivered during the first of two cycles of gemcitabine, given at the current recommended dose when used alone. We have concluded that 42 Gy, delivered to the primary tumor in 2.8-Gy fractions, does not exceed the MTD as defined in the protocol. However, the occurrence of dose-limiting gastrointestinal toxicity in two patients at this final dose level suggests that further dose escalation may result in intolerable toxicity. We have elected not to investigate a higher dose on the basis of this observation and the potential for late toxicity. The concern for late toxicity is based on radiobiologic data that indicate an increasing risk for late toxicity as the fraction size increases, yet late toxicity is also a function of the total dose. The linear quadratic model, which can account for both fraction size and total dose, indicates that 42 Gy in 2.8-Gy fractions is biologically equivalent (with regard to late effects) to 50.4 Gy in 1.8-Gy fractions, which is a relatively standard dose and fractionation schedule used in the management of patients with unresectable pancreatic cancer. Additional accrual at the 39-Gy dose level was not pursued because 42 Gy did not exceed the MTD, as defined in the protocol. As a result, our recommendation of 36 Gy in 2.4-Gy fractions as a phase II dose is based on experience in only three patients at each of the two prior dose levels (36 and 39 Gy). This recommendation is a conservative one, based again on the concern for late effects, particularly considering lack of long-term follow-up of the patients treated at 39 Gy. The 36-Gy dose is biologically equivalent to approximately 41.4 Gy in 1.8-Gy fractions with regard to late effects. This recommended dose is also accompanied by a note of caution regarding the radiation field size, which is defined much differently than radiation field sizes used in more conventional combined-modality regimens. The rate of myelosuppression with our regimen was quite similar to that observed in the gemcitabine arm of the phase III randomized trial comparing gemcitabine with 5-FU,7 suggesting no greater hematologic toxicity with the addition of radiation therapy. This is supported further by data that indicate no difference in the gemcitabine dose received during the first and second cycles of protocol therapy. Our data on local and regional control suggest that the reduction in radiation dose and field size required in this regimen has not resulted in excess failures at these sites. Thus further development of this regimen is warranted to investigate our hypothesis that this novel chemoradiotherapy approach, which emphasizes systemic therapy, may improve survival for patients with pancreatic cancer. Gemcitabine-based chemoradiation approaches in pancreatic cancer patients have been investigated by several groups and remain under investigation by others. In a trial reported by Blackstock et al,12 using gemcitabine twice weekly in an attempt to maximize radiosensitization, both hematologic and gastrointestinal toxicity were dose-limiting. Hematologic toxicity noted in this trial at 60 mg/m2 is near the MTD of 65 mg/m2 when gemcitabine alone is delivered twice weekly.13 It is unlikely that radiation therapy influenced hematologic toxicity, as observed in the current study. Gastrointestinal toxicity with twice weekly administration may have been influenced by the schedule of gemcitabine administration, with increased radiosensitization of normal tissues. However, the inclusion of prophylactic nodal basins in the treatment volume, resulting in a large volume of normal tissue irradiated, may have been a more critical factor. Investigators at M.D. Anderson Cancer Center have reported significant gastrointestinal toxicity when weekly gemcitabine is delivered at doses more than 350 mg/m2 with concurrent rapid fractionation of 30 Gy in 3-Gy fractions.14 Our results suggest that the toxicity encountered in this trial is unlikely to be related to the increased dose per fraction. Gemcitabine was delivered on the Friday before initiation of radiation and continued weekly at the end of each radiation treatment week. This schedule was based on preclinical data indicating more selective tumor radiosensitization with exposure at least 24 hours before radiation and may have contributed to the gastrointestinal toxicity observed despite modest doses of gemcitabine and radiation.15 The treatment volumes previously reported with rapid fractionation covered the primary tumor with a 3- to 5-cm margin, as well as porta hepatis and celiac axis lymph nodes.16,17 The use of similar fields in the gemcitabine trial may be implicated in the toxicity encountered. In a recent Eastern Cooperative Oncology Group trial, the use of gemcitabine with concurrent PVI 5-FU and radiation therapy was investigated, with weekly gemcitabine doses ranging from 50 to 100 mg/m2.18 Again, the radiation treatment volume included nodes at risk for occult metastases. Three of seven patients on the trial experienced gastrointestinal DLT. In two patients, this occurred after completion of radiotherapy (59.4 Gy in 1.8-Gy fractions). In this trial then, the delivery of concurrent PVI 5-FU, a relatively high-dose of radiation therapy, and the treatment of volumes including lymph nodes at risk may have contributed to the toxicity observed. It is apparent that a number of variables must be critically assessed in the design and interpretation of clinical trials investigating gemcitabine-based radiosensitization. Relatively large treatment volumes were used in each of the three trials described above, all of which reported substantial gastrointestinal toxicity. This contrasts with the relative lack of gastrointestinal toxicity in our experience, using a more conformal approach and exclusion of prophylactic nodal irradiation. Thus the radiation treatment volume is perhaps the most critical variable influencing gastrointestinal toxicity in gemcitabine-based chemoradiotherapy regimens. The objective responses observed on this trial were encouraging, particularly when accompanied by weight gain or decrease in tumor-related symptoms. We have elected to include patients who responded after additional chemotherapy in the overall response rate, as this may reflect a delayed response from radiation therapy. The objective response rate after gemcitabine alone reported in the investigational new drug treatment program was 12% (95% CI, 10% to 14%).19 Although response was not the primary end point in the trial, it is reasonable to conclude that the addition of radiation resulted in an improved response rate, given the overall response rate of 30% (95% CI, 17% to 48%). Patterns of failure are difficult to document in patients with pancreatic cancer and often the site of first failure is the only site reported. We attempted to score all sites of failure, considering the possibility that excess local and regional failures could result from a reduction of the radiation dose and a smaller treatment volume. Local failures occurred relatively infrequently. It is quite possible that clinical radiosensitization occurred, considering the reduction in the dose of radiation therapy. Regional adenopathy was also an infrequent event, occurring in only three of 37 patients. This is perhaps the most significant finding with respect to patterns of failure, as it suggests that the elimination of prophylactic nodal irradiation during administration of full-dose gemcitabine does not result in excessive failure at this site. Given that CT scans were not mandated after the scan obtained at week 9, it is possible that additional local or regional failures occurred, particularly in patients who developed metastatic disease. Of the patients with documented failure, the majority failed or progressed at distant sites. This suggests that further emphases on systemic therapy would be warranted. Our current ongoing trial is investigating incorporation of cisplatin into the protocol reported, based on data suggesting greater systemic activity with gemcitabine and cisplatin compared with prior experience with single-agent gemcitabine20,21 A median survival of 11.6 months in this population of patients compares favorably with data indicating a 10-month median survival after 5-FU–based chemoradiotherapy and a 6.6-month median survival after gemcitabine alone in patients with locally advanced disease and no evidence of metastases.1,19 The lack of a survival difference between patients with or without metastases in our trial may have resulted from selection of patients with a minimal burden of metastatic disease. Yet the outcome among these patients suggests that radiation therapy may have a role in patients such as these, particularly among those with large symptomatic primary tumors. This gemcitabine-based chemoradiotherapy regimen is clearly tolerable, a finding in contrast with previously reported approaches.12,14,18 The response rate observed, control of the tumor at the primary site, and the lack of excess failure at the regional nodes suggest that this approach, which attempts to maximize systemic disease control, should be studied in the phase II setting. Survival data from such a trial may then support investigation in a phase III trial.
Supported in part by Eli Lilly and Co, Indianapolis, IN.
Presented in part at the Forty-Second Annual Meeting of the American Society for Therapeutic Radiology and Oncology, Boston, MA, October 21-25, 2000, and the Thirty-Fifth Annual Meeting of the American Society of Clinical Oncology, Atlanta, GA, May 15-18, 1999.
1. Gastrointestinal Tumor Study Group: Therapy of locally unresectable pancreatic carcinoma: A randomized comparison of high dose (6000 rads) radiation alone, moderate dose radiation (4000 rads + 5-fluorouracil), and high dose radiation + 5-fluorouracil. Cancer 48: 1705-1710, 1981[Medline] 2. Mohiuddin M, Regine WF, Stevens J, et al: Combined intraoperative radiation and perioperative chemotherapy for unresectable cancers of the pancreas. J Clin Oncol 13: 2764-2768, 1995[Abstract] 3. Garton GR, Gunderson LL, Nagorney DM, et al: High-dose preoperative external beam and intraoperative irradiation for locally advanced pancreatic cancer. Int J Radiat Oncol Biol Phys 27: 1153-1157, 1993[Medline] 4. Tepper JE, Noyes D, Krall JM, et al: Intraoperative radiation therapy of pancreatic carcinoma: A report of RTOG-8505. Int J Radiat Oncol Biol Phys 21: 1145-1149, 1991[Medline] 5. Poen JC, Collins HL, Niederhuber JE, et al: Chemo-radiotherapy for localized pancreatic cancer: Increased dose intensity and reduced acute toxicity with concomitant radiotherapy and protracted venous infusion 5-fluorouracil. Int J Radiat Oncol Biol Phys 40: 93-99, 1998[Medline]
6.
Whittington R, Neuberg D, Tester WJ, et al: Protracted intravenous fluorouracil infusion with radiation therapy in the management of localized pancreaticobiliary carcinoma: A phase I Eastern Cooperative Oncology Group Trial. J Clin Oncol 13: 227-232, 1995
7.
Burris HA, Moore MJ, Andersen J, et al: Improvements in survival and clinical benefit with gemcitabine as first line therapy for patients with advanced pancreas cancer: A randomized trial. J Clin Oncol 15: 2403-2413, 1997 8. Lawrence TS, Chang EY, Hahn TM, et al: Radiosensitization of pancreatic cancer cells by 2',2'-difluoro-2'-deoxycytidine. Int J Radiat Oncol Biol Phys 34: 867-872, 1996[Medline] 9. National Comprehensive Cancer Network: NCCN practice guidelines for pancreatic cancer. Oncology 11: 41-55, 1997 10. Robertson JM, Marsh L, TenHaken RK, et al: The clinical application of a non-axial treatment plan for pancreatic and biliary malignancies. Radiother Oncol 24: 198-200, 1992[Medline] 11. Miller AB, Hoogstraten B, Staquet M, et al: Reporting results of cancer treatment. Cancer 47: 207-214, 1981[Medline]
12.
Blackstock AW, Bernard SA, Richards F, et al: Phase I trial of twice-weekly gemcitabine and concurrent radiation in patients with advanced pancreatic cancer. J Clin Oncol 17: 2208-2212, 1999 13. Poplin EA, Corbett T, Flaherty L, et al: Difluorodeoxycytidine (dFdC)–gemcitabine: A phase I study. Invest New Drugs 10: 165-170, 1992[Medline]
14.
Wolff RA, Evans DB, Gravel DM, et al: Phase I trial of gemcitabine combined with radiation for the treatment of locally advanced pancreatic adenocarcinoma. Clin Cancer Res 7: 2246-2253, 2001 15. Mason KA, Milas L, Hunter NR, et al: Maximizing therapeutic gain with gemcitabine and fractionated radiation. Int J Radiat Oncol Biol Phys 44: 1125-1135, 1999[Medline] 16. Evans DB, Rich TA, Byrd DR, et al: Preoperative chemoradiation and pancreaticoduodenectomy for adenocarcinoma of the pancreas. Arch Surg 127: 1335-1339, 1992[Abstract]
17.
Pisters PW, Abbruzzese JL, Janjan NA, et al: Rapid-fractionation preoperative chemoradiation, pancreaticoduodenectomy, and intraoperative radiation therapy for resectable pancreatic adenocarcinoma. J Clin Oncol 16: 3843-3850, 1998
18.
Talamonti MS, Catalano PJ, Vaughn DJ, et al: Eastern Cooperative Oncology Group phase I trial of protracted venous infusion fluorouracil plus weekly gemcitabine with concurrent radiation therapy in patients with locally advanced pancreas cancer: A regimen with unexpected early toxicity. J Clin Oncol 18: 3384-3389, 2000 19. Storniolo AM, Enas NH, Brown CA, et al: An investigational new drug treatment program for patients with gemcitabine: Results for over 3000 patients with pancreatic carcinoma. Cancer 85: 1261-1268, 1999[Medline] 20. Philip P, Zalupski M, Vaitkevicius VK, et al: Phase II study of gemcitabine and cisplatin in advanced or metastatic pancreatic cancer. Proc Am Soc Clin Oncol 18: 1053, 1999 21. Zalupski MM, Lawrence TS, Hejna G, et al: A phase I trial of cisplatin plus gemcitabine and radiation therapy in pancreatic cancer. Proc Am Soc Clin Oncol 20: 144b, 2001 (abstr 2327) Submitted January 26, 2001; accepted July 19, 2001. This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
18 years, Zubrod performance status of 
2, estimated life expectancy of at least 12 weeks, and adequate organ function. Patients with a history of prior upper abdominal radiation therapy or chemotherapy were ineligible, excepting patients who may have received one cycle of gemcitabine before study entry. In this case, a 2-week delay was required between the last gemcitabine dose and initiation of protocol treatment. The trial was approved by the institutional review board of the University of Michigan Medical School and William Beaumont Hospital Human Investigation Committee. Written informed consent was obtained from all patients before initiation of therapy. 
