Adjuvant Therapy for Pancreatic Cancer: To Treat or Not To Treat?
Adjuvant Therapy for Pancreatic Cancer: To Treat or Not To Treat?
Despite attempted curative resection of localized adenocarcinoma of the pancreas, most patients experience a recurrence and die of their disease. The Gastrointestinal Tumor Study Group, European Organisation for Research and Treatment of Cancer, and European Study Group for Pancreatic Cancer trials have suggested the benefit of adjuvant therapy. However, the relatively few randomized trials available have not established a definite standard of care due to study limitations. Although these trials, and the recently published Charité Onkologie (CONKO)-001 trial, have shown a definite advantage of adjuvant chemotherapy, the most effective chemotherapy and the role of radiation therapy remain unclear. This review will discuss the data available from reported trials of adjuvant and neoadjuvant therapy in pancreatic cancer, address the issues leading to the ongoing controversies, and consider future directions for clinical trials.
Surgical resection remains the only curative therapy for localized pancreatic cancer. Unfortunately, less than 20% of patients present with potentially resectable disease. Even among patients who undergo successful resection, the rate of recurrence is around 90%, and local recurrence occurs in 50% to 80%, with a 5-year overall survival of only 10% to 15%. Due to the high recurrence rate and poor survival of resected patients, adjuvant therapy is a logical treatment option.
The first randomized trial providing evidence in favor of adjuvant therapy was conducted by the Gastrointestinal Tumor Study Group (GITSG). In this trial, Kalser et al compared observation to bolus fluorouracil (5-FU) plus split-course radiation followed by 2 years of 5-FU. The study was closed early due to slow accrual with a total of 43 evaluable patients, and an interim analysis showed a statistically significant improvement in survival. An improvement in median survival of 20 vs 11 months (P = .035), and in 2-year survival of 42% vs 15%, was shown for the adjuvant chemoradiation group. This trial was the first to show a survival advantage, but the study population was small, and it used a radiation regimen that is now felt to be inferior. Additionally, since all patients received both chemoradiation and chemotherapy, the relative contribution of each modality could not be assessed. Nonetheless, as the first such report, this study started the use of combined chemoradiation as an American standard for adjuvant therapy.
The European Organisation for Research and Treatment of Cancer (EORTC) compared observation to adjuvant infusional 5-FU and split-course radiation in a larger group of 218 patients. This trial included patients with periampullary cancers, and only 114 of the 207 evaluable patients had pancreatic head cancers, but the two groups were stratified before randomization. Among all patients, the investigators found a non-statistically significant difference in 2-year survival between the treatment and control groups—51% (95% CI = 41%-61%) vs 41% (95% CI = 31%-51%), respectively—and a relative risk of death of 0.8 (95% CI = 0.6-1.1). In the analysis of pancreatic head cancer patients, 2-year survival was 37% (CI = 24%-50%) for the treatment arm vs 23% (CI = 11%-35%) for the control arm. Although there appears to be a trend favoring the treatment arm, no conclusion about the benefit of adjuvant chemoradiation in pancreatic cancer can be made from this trial.
While the EORTC trial studied only combined chemoradiation, the GITSG trial included both chemoradiation and subsequent chemotherapy but could not differentiate the effect of these two modalities. The European Study Group for Pancreatic Cancer (ESPAC) group attempted to address adjuvant therapy with chemotherapy, radiation, and chemoradiation. In the ESPAC-1 trial the investigators began with a 2X2 factorial design with one randomization to chemoradiation or no chemoradiation and the other to chemotherapy vs no chemotherapy. This resulted in four treatment groups: observation, concurrent chemoradiation, chemotherapy alone, and chemoradiation followed by chemotherapy (Figure 1). The study was powered to compare chemoradiation vs no chemoradiation and chemotherapy vs no chemotherapy.
In addition to this 2X2 factorial design, an additional 256 patients were allocated to one of two randomizations, chemoradiation vs observation, or chemotherapy vs observation. According to the authors, this option was allowed because some study centers may not have had access to both modalities. The treating physician was allowed to choose the randomization for each patient, although it was not intended for these patients to be part of the planned primary analysis.
In the first ESPAC-1 report, all three randomization methods were combined in the data analysis. The 175 patients receiving chemoradiation in the 2X2 factorial and the chemoradiation randomization were compared to the 178 patients who did not receive chemoradiation (ie, the observation and chemotherapy-only groups). There was no difference in 2-year survival between the two groups—15.5 vs 16.1 months, respectively. Similarly, in the 2X2 design and the chemotherapy randomization groups, the 238 patients who received chemotherapy alone were compared to the 235 patients who did not (observation and chemoradiation). This comparison did show a statistically significant improvement in 2-year survival—19.7 months (95% CI = 16.4-22.4) vs 14 months (95% CI = 11.9-6.5).
After a median follow-up of 47 months, data were reported for the 289 patients from the primary analysis group involved in the 2X2 randomization design. The 2-year survival for patients who received "chemotherapy alone" (both the chemotherapy and chemoradiation-followed-by-chemotherapy groups) was 40%, whereas the survival for those who did not receive chemotherapy alone (observation and chemoradiation groups) was 30%. Five-year survival also improved at 21% vs 8%, respectively (no P value given). Conversely, chemoradiation appeared to be detrimental, as the 2-year survival in those patients who received chemoradiation was 29% compared to 41% in those who did not receive any radiation, with a hazard ratio for death of 1.28 (CI = 0.99-1.66, P = .05).
The authors of the ESPAC-1 trial concluded that all patients should receive adjuvant chemotherapy. However, several issues must be considered in interpreting these results. The statistical design was written originally to evaluate a primary endpoint of 2-year survival for margin-negative patients. The final results for this endpoint were never reported. Appropriately, no comparison was made for each of the four individual treatment groups, as ESPAC-1 was not powered for this. Although the addition of the other randomizations adds complexity to the interpretation of the results, the data from the single randomization groups were appropriately not included in the final analysis and paper.
The ESPAC-1 authors also suggested that chemoradiation is ineffective. However, these results are relevant to the split-course radiation used in this trial and cannot be extrapolated to more modern chemoradiation regimens. Use of infusional 5-FU regimens and single-course radiation without breaks may provide benefit, and these strategies have not yet been compared to non-radiation-containing arms in prospective, randomized trials. Additional concerns for the chemoradiation question in ESPAC-1 have been raised. First, the doses of radiation varied between sites despite a recommendation for 40 Gy. Second, the lack of central review for radiation ports is of significant importance.
Recently, investigators (including the authors of ESPAC-1) performed a meta-analysis to address the issues of adjuvant chemoradiation and adjuvant chemotherapy for pancreatic cancer separately. This analysis showed a benefit for chemotherapy with a pooled hazard ratio of 0.75 (CI = 0.64-0.90), and no effect for chemoradiation (HR = 1.09, CI = 0.89-1.32). Analysis of prognostic subgroups showed a potential benefit for chemoradiation in patients with positive margins, with a hazard ratio of 0.62 (but the confidence interval was wide).
The majority of patients in the meta-analysis were from the ESPAC-1 trial, including those randomized in the 2X2 design as well as those in the separate randomization groups. In addition, subgroups of three other trials were used in this analysis. ESPAC-1 weighed significantly in determining the results, and the use of subgroups may not be valid for meta-analysis. A meta-analysis is designed only as a tool for hypothesis generation, and based on the limitations of this report, the results should be interpreted with caution.