Author: JP Neoptolemos, et al. [ESPAC-1]
Source: New Engl J of Medicine 2004, 350(12);1200-10
This publication is the 2004 update to the initial report of the ESPAC-1 trial from Europe which was intended to be a randomized study examining whether patients with resected pancreatic cancer benefit from postoperative adjuvant therapy in the form of chemotherapy, radiotherapy, or a combination of the two.
Only 10-15% of patients with pancreatic cancer present with localized disease amenable to surgical resection. While patients with unresectable disease have a dismal prognosis, resected patients still have an unacceptable rate of mortality from their disease. Numerous studies dating back to the 1960s have examined the role of adjuvant therapy following surgical resection to improve survival, however conclusions remain mixed. At this point in time, a benefit from additional postoperative therapy remains controversial. The ESPAC-1 trial was designed to be a large multi-national European effort to enroll hundreds of patients into a randomized trial to help answer the question of whether adjuvant therapy can benefit patients following surgical resection. Unfortunately, there are severe design flaws in this trial which preclude scientific conclusions from the data presented in both publications.
The study intended to randomize patients into one of 4 groups following surgical resection: (A) Observation, (B) chemoradiotherapy, (C) chemotherapy, or (D) chemoradiotherapy followed by maintenance chemotherapy. Approximately half of the patients were randomized via a 2x2 factorial process, and the other half were allowed to be selected by participating physicians who had a preference for a specific form of adjuvant therapy. This latter half of patients were allowed receive background therapy at the discretion of their physician. The initial 2001 publication reported on the results of all patients, while this latest publication reports only on the 289 patients who were randomized by the 2x2 factorial design. Unfortunately, the patients randomized via the 2x2 factorial design had already been predisposed to the bias of physician selection.
The 4 above groups were then categorized into (1) Chemoradiotherapy, (2) No Chemoradiotherapy, (3) Chemotherapy, and (4) No Chemotherapy. The Chemoradiotherapy category consisted of groups B & D; the No Chemoradiotherapy category consisted of groups A & C; the Chemotherapy category consisted of groups C & D; and the No Chemotherapy category consisted of groups A & D.
Radiotherapy technique mirrored that used in the 1970s GITSG trial: 40 Gy split course of radiation consisting of 2 weeks of treatment followed by a 2 week break followed by another 2 weeks of radiation. Chemotherapy consisted of bolus 5FU with Leucovorin when given alone, and bolus 5FU only when given concurrently with radiation.
Data from the 2x2 randomization was included in this updated analysis. The survival data from each of the 4 randomization groups, although underpowered for analysis, revealed 5 year survival rates of 11%, 7%, 29%, and 13% for postoperative observation, chemoradiotherapy, chemotherapy alone, and chemoradiotherapy followed by maintenance chemotherapy, respectively.
Using the categorization described above, the Chemoradiotherapy, No Chemoradioatherapy, Chemothearpy, and No Chemotherapy groups each had a total of 144, 145, 142, and 148 patients for 2x2 comparison. The time to recurrence and 1 year disease-free survival for chemoradiotherapy vs no chemoradiotherapy was 10.7 months and 46% vs 15.2 months and 58%, respectively. When comparing chemotherapy vs no chemotherapy, the time to recurrence and 1 year disease-free survival was 15.3 months and 58% vs 9.4 months and 43%, respectively.
The published analysis of survival data for Chemoradiotherapy vs No Chemoradiotherapy groups revealed a median, 2-year, and 5 year survival rate of 15.9 mo, 29%, and 10% vs 17.9, 41%, and 20%, respectively. Examining Chemotherapy vs No Chemotherapy, the median, 2-year, and 5 year survival rates were 20.1 months, 40%, and 21% vs 15.5 months, 30%, and 8%, respeively.
20% of patients were found to have positive margins. Margins status had been previously analyzed in a separate publication which confirmed the adverse prognostic effect with positive margins, but were able to show that patients with negative margins still had similar benefit and lack of benefit from chemotherapy or chemoradiotherapy, respectively.
The study design introduced physician bias, and an antiquated form of radiotherapy was used. While purported to be a randomized trial, this trial in effect was a non-randomized prospective trial with selection biases. The physician could decide to randomize to only one form of treatment, or give background adjuvant therapy at their discretion and as summarized by Tepper, to say this could lead to bias is an understatement. The separate treatment groups were underpowered for individual analyses, and thus a complicated scheme was introduced categorizing 2 groups to compare to 2 other groups when examining chemoradiotherapy vs no chemoradiotherapy, or chemotherapy vs no chemotherapy.
The authors conclude that adjuvant chemotherapy can improve survival, while combining radiation with chemotherapy actually is harmful. Based on the flaws in this study, it is not clear what conclusions can be drawn from the data available. Numerous editorials have been written about this paper both in favor and in defense of the study, and thus the results at this point remain controversial.