Surgical therapy currently offers the only potential cure for pancreatic adenocarcinoma. Surgical morbidity and mortality have decreased dramatically in recent years; the perioperative mortality associated with pancreaticoduodenectomy in major centers is approximately 1 percent. However, only a few patients present with tumors that are amenable to resection, and even after resection of a seemingly localized neoplasm, long-term survival is poor.1 Since many medical centers now have the capacity to resect pancreatic cancer safely, it is increasingly important to identify effective postoperative (adjuvant) therapy if we are to achieve long-term success in treating this disease.

The potential benefit of adjuvant therapy after resection of pancreatic cancer was first recognized by the randomized trial conducted by the Gastrointestinal Tumor Study Group (GITSG) almost 20 years ago. Since then, many reports from single institutions have shown a benefit of adjuvant treatment. Among the few randomized trials is the comparison of chemoradiotherapy with observation after the resection of pancreatic cancer, conducted by the European Organization for Research and Treatment of Cancer (EORTC). This study was statistically underpowered, but a trend toward improved survival was seen in the chemoradiotherapy group. Despite slowly emerging evidence of a benefit, the role of postoperative therapy in the management of pancreatic cancer is inadequately defined.

In this issue of the Journal, Neoptolemos and his colleagues report an updated and improved analysis of the European Study Group for Pancreatic Cancer (ESPAC-1) trial, the largest randomized trial of adjuvant therapy for pancreatic cancer reported to date. The original report from this group6 conveyed interim results for 541 patients in a randomized trial. This study combined the results of a trial with a two-by-two factorial design with those of two additional trials in which patients were randomly assigned to either chemoradiotherapy or no chemoradiotherapy and to either chemotherapy or no chemotherapy. The median follow-up at that time was 10 months. Although the results demonstrated a survival benefit associated with adjuvant chemotherapy, but not with chemoradiotherapy, the study was criticized, in part on the grounds that there was selection bias in the pooled data.

The current study represents an attempt to correct some of the limitations of the previous report by examining only the 289 patients who underwent strict randomization according to the factorial design and reporting results at a median follow-up of 47 months. Patients were randomly assigned to one of four groups. Although the groups were not analyzed separately, the design seemingly made it possible to determine the independent effect of each treatment on survival. This analysis led the authors to conclude that postoperative chemotherapy with fluorouracil plus leucovorin conferred a benefit in terms of survival, whereas postoperative chemoradiotherapy had a deleterious effect on survival.

The main advantage of a two-by-two factorial design is that it permits investigators to explore two potentially independent effects within a single study. The chemoradiotherapy regimen included a short course of radiosensitizing fluorouracil, which can be considered different from and perhaps independent of subsequent chemotherapy with a prolonged course of fluorouracil plus leucovorin. The design of such a trial provides the opportunity to analyze the main effect of each therapy and to identify any interaction or synergy between the two therapies. In the study by Neoptolemos et al., however, the two treatments — chemoradiotherapy and chemotherapy — were administered consecutively, with the first treatment probably influencing compliance with the second for those assigned to both. Indeed, of 147 patients assigned to receive chemotherapy with or without chemoradiotherapy, 33 percent of the 122 for whom treatment details were available did not complete the chemotherapy regimen and 17 percent received no chemotherapy. Details regarding the causes of nonadherence, including whether previous chemoradiothrapy affected the rate of completion of subsequent chemotherapy, are not presented in the article. The high rate of nonadherence and the potential for bias arouse concern regarding the validity of such an analysis and therefore its conclusions. Intention-to-treat analysis does not remove the potential bias reintroduced into the randomization structure by the sequential-therapy design.

One lesson to be learned from this well-intentioned study is that a straightforward analysis driven by the factorial structure is not always as simple as it seems. Because of confounding by the sequential nature of the two therapies and the high rate of nonadherence, the two-by-two design obligates investigators to provide detailed data on the survival of each of the four treatment groups.

The authors report that the trial was not powered sufficiently for them to perform statistical comparisons of the four treatment groups, but the differences in survival between individual groups do shed some light on the probable reason for the poorer survival among those treated with chemoradiotherapy. Patients who received chemoradiotherapy alone had a worse median survival than those who received no adjuvant therapy. Unless one believes that disease progression was somehow promoted by therapy, it is highly likely that the deleterious effect of chemoradiotherapy was indeed due to treatment-related toxic effects.

How should the results of this study influence future trials of adjuvant therapy for pancreatic cancer? Some may feel that the important question is whether future studies should include radiation or whether, instead, more rigorous specifications, support guidelines, and quality-assurance monitoring are necessary before this therapy is included in a trial of adjuvant therapy. Others may ask whether it is appropriate to include groups that do not receive radiation, given that other results support the role of radiation. The ongoing and maturing trials in Europe and the United States reflect general opinion regarding the future role of adjuvant radiation therapy. The ESPAC-3 trial, now in progress, does not include any radiation; rather, it is designed to compare three groups after resection of pancreatic cancer: patients receiving fluorouracil plus leucovorin, those receiving gemcitabine, and those receiving no treatment. Results of the closed Radiation Therapy Oncology Group trial 97-04 are expected later this year. In this trial, chemoradiotherapy was given to all patients with resected pancreatic cancer. Patients were randomly assigned to receive either gemcitabine or infusional fluorouracil given before and after chemoradiotherapy.

Current single-institution and phase 2 studies are exploring the role of newer chemotherapeutic and biologic agents. Nukui  reported data on 33 patients who received chemoradiotherapy, which included fluorouracil, cisplatin, and interferon alfa, followed by infusional fluorouracil. Although associated with somewhat increased toxicity, this regimen resulted in a remarkable 2-year survival rate of 84 percent and a median survival of 45 months. These results await confirmation in an ongoing phase 2 trial by the American College of Surgeons Oncology Group. Newer biologic agents, including farnesyl transferase inhibitors and monoclonal antibodies such as trastuzumab and cetuximab, are currently being tested in patients with metastatic pancreatic cancer.

As these studies evolve, promising new therapies will certainly make their way into adjuvant-therapy trials. In the future, decisions about adjuvant therapy will probably be influenced by improved methods for the assessment of the risk of recurrence, by the availability of more accurate surgical staging methods, and by the application of molecular diagnostic techniques. Many remain hopeful that advances in systemic and locoregional therapies, along with improvements in risk assessment and early detection of pancreatic cancer, will result in an optimism heretofore not seen among patients and physicians concerned with this disease.