The relatively high rate of both locoregional and distant recurrence following surgery for pancreatic cancer makes a strong case for effective adjuvant therapy [40
]. RCTs of CRT are limited, and the available data are boosted by some phase II and single-institution studies. They are summarized in [42
Adjuvant chemoradiation in pancreatic adenocarcinoma
The first prospective, multicenter trial of CRT versus observation alone was performed by the GITSG [42
]. Resected pancreatic cancer patients with R0 margins were assigned to receive either split-course radiotherapy over 6 weeks with a 2-week gap in between, with concurrent 5-FU on week 1 and week 5 followed by maintenance 5-FU for 2 years or until progression or no active treatment. In 1974–1982, only 49 patients were randomized. At an interim analysis, patients in the CRT arm had a significantly longer median survival time (21 months versus 11 months). A further 32 patients were added to the treatment cohort in a nonrandomized fashion following the interim analysis, and the final analysis showed a median survival time of 18 months with 2- and 5-year survival rates of 46% and 17%, respectively [49
]. Following that trial, adjuvant CRT became the standard of care in the U.S. The study, however, has been criticized in other quarters for its poor accrual, low statistical power, suboptimal radiotherapy schedule, lack of radiotherapy quality assurance, and noncompliance with maintenance chemotherapy in 75% of patients.
The European Organization for Research and Treatment of Cancer (EORTC) conducted a similar study of CRT versus observation in Europe between 1987 and 1995 [43
]. The radiation schedule was similar but the 5-FU was delivered as an infusion and there was no maintenance chemotherapy. It included pancreatic and periampullary cancers and both R0 and R1 resections, but did not prestratify for primary site or resection margin status. The trial did not show any significant benefit in terms overall survival in the whole population or in patients with pancreatic head cancer. The statistical analysis of that trial has been criticized, and it could have given a significant result if the design was more appropriate [50
The European Study Group for Pancreatic Cancer (ESPAC) conducted the largest phase III RCT in this setting between 1994 and 2000. Five hundred forty-one patients were randomized to: (a) chemotherapy versus observation, (b) CRT versus observation, and (c) a 2 × 2 factorial design of observation versus chemotherapy versus CRT versus CRT plus maintenance chemotherapy. The radiotherapy schedule was similar to that used in the EORTC study and the chemotherapy agent was bolus 5-FU. Both R0 and R1 patients were included. At an early intent-to-treat analysis at 10 months, there was a statistically significant survival benefit for patients receiving chemotherapy (median survival time, 19.7 months versus 14 months; p
= .0005) but no benefit for patients treated with CRT (median survival time, 15.5 months versus 16.1 months; p
= .24) [44
]. The mature results of ESPAC-1 [45
] with analysis restricted to the 2 × 2 arm of the study showed a significant 5-year survival benefit for chemotherapy versus no chemotherapy (21% versus 8%; p
= .009), but no benefit for CRT versus no CRT (10% versus 20%; p
= .05). The conclusion from that trial was that adjuvant chemotherapy significantly improved survival, whereas CRT had a detrimental effect on survival because it delayed systemic chemotherapy. The results generated substantial controversy and the trial was criticized because of a suboptimal radiotherapy schedule, lack of central radiotherapy quality assurance, wide variation in the radiotherapy doses employed, in violation of the protocol, and the allowance of background therapy with chemotherapy or CRT prior to randomization, which could all potentially influence the final analysis.
A subsequent meta-analysis of adjuvant therapy by the Pancreatic Cancer Meta-analysis Group (PCMG) in 2005 looked at individual patient data from five randomized studies of chemotherapy and CRT along with previously unpublished results of the ESPAC-1 study [51
]. They concluded that: (a) chemotherapy alone reduced the risk for death by 25% (hazard ratio [HR], 0.75; confidence interval [CI], 0.64–0.90; stratified p
= .001), (b) CRT had no significant impact (HR, 1.09; CI, 0.89–1.32; stratified p
= .43), and (c) subgroup analyses showed CRT as more effective than chemotherapy in patients with R1 resections. However, that meta-analysis was heavily influenced by ESPAC-1 data.
A further PCMG meta-analysis looking at the influence of resection margin status and treatment on survival suggested that resection margin involvement was not a significant factor for survival (HR, 1.10; CI, 0.94–1.29; p
= .24) [52
]. The 2- and 5-year survival rates were 33% and 16% for R0 and 29% and 15% for R1 patients, respectively. CRT in R1 patients resulted in a 28% lower risk for death (HR, 0.72; CI, 0.47–1.10) and there was a 19% higher risk for death in R0 patients (HR, 1.19; CI, 0.95–1.49). Chemotherapy, on the other hand, resulted in a 4% higher risk for death in R1 patients (HR, 1.04; CI, 0.78–1.40) and a 35% lower risk for death in R0 patients (HR, 0.65; CI, 0.53–0.80).
The latest trial of CRT (Radiation Therapy Oncology Group trial 97–04) was conducted between 1998 and 2002 [46
]. The analysis was conducted on 442 of 538 patients randomized between 3 weeks before and 3 months after CRT with 5-FU and 3 weeks before and 3 months after CRT with gemcitabine. The CRT part in both arms delivered 50.4 Gy in 28 fractions, with concurrent 5-FU as a 250-mg/m2
per day continuous infusion. Only 5% of the patients had an unacceptable deviation from protocol. Patients were stratified for surgical margin, tumor diameter, and nodal status. At the final analysis, 381 patients with pancreatic head tumors only had a significant benefit from gemcitabine in terms of the median survival time and 3-year survival rate (20.6 months versus 16.9 months and 32% versus 21%, respectively). There was no significant difference when tumors of the body and tail were included as well.
The ESPAC group recently reported data from ESPAC-3 in their latest abstract, showing equivalence of gemcitabine and 5-FU plus leucovorin as adjuvant therapy, with a better safety profile in favor of gemcitabine [53
]. ESPAC-4 has now been launched comparing gemcitabine with gemcitabine plus capecitabine in the adjuvant setting because the assessment of the ESPAC group is that CRT offers no benefit in this setting. Nevertheless, the issue of the optimal treatment of patients with positive resection margins is still far from clear.
In light of the above findings, it is difficult to formulate a “one-size-fits-all” strategy in the adjuvant setting for pancreatic cancer. Good quality trials are still needed, targeting surgical subgroups, especially because more aggressive surgery of “borderline” cases will lead to a greater number of R1 resections.
The advent of biologicals is interesting, but it is as yet difficult to see where they fit in the combination radiotherapy and adjuvant settings, given disappointing results in APC patients to date.