Preoperative Modified FOLFIRINOX Treatment Followed by Capecitabine-Based Chemoradiation for Borderline Resectable Pancreatic Cancer: Alliance for Clinical Trials in Oncology Trial A021101

Abstract

Importance: Although consensus statements support the preoperative treatment of borderline resectable pancreatic cancer, no prospective, quality-controlled, multicenter studies of this strategy have been conducted. Existing studies are retrospective and confounded by heterogeneity in patients studied, therapeutic algorithms used, and outcomes reported.

Objective: To determine the feasibility of conducting studies of multimodality therapy for borderline resectable pancreatic cancer in the cooperative group setting.

Design, setting, and participants: A prospective, multicenter, single-arm trial of a multimodality treatment regimen administered within a study framework using centralized quality control with the cooperation of 14 member institutions of the National Clinical Trials Network. Twenty-nine patients with biopsy-confirmed pancreatic cancer preregistered, and 23 patients with tumors who met centrally reviewed radiographic criteria registered. Twenty-two patients initiated therapy (median age, 64 years [range, 50-76 years]; 55% female). Patients registered between May 29, 2013, and February 7, 2014.

Interventions: Patients received modified FOLFIRINOX treatment (85 mg/m2 of oxaliplatin, 180 mg/m2 of irinotecan hydrochloride, 400 mg/m2 of leucovorin calcium, and then 2400 mg/m2 of 5-fluorouracil for 4 cycles) followed by 5.5 weeks of external-beam radiation (50.4 Gy delivered in 28 daily fractions) with capecitabine (825 mg/m2 orally twice daily) prior to pancreatectomy.

Main outcomes and measures: Feasibility, defined by the accrual rate, the safety of the preoperative regimen, and the pancreatectomy rate.

Results: The accrual rate of 2.6 patients per month was superior to the anticipated rate. Although 14 of the 22 patients (64% [95% CI, 41%-83%]) had grade 3 or higher adverse events, 15 of the 22 patients (68% [95% CI, 49%-88%]) underwent pancreatectomy. Of these 15 patients, 12 (80%) required vascular resection, 14 (93%) had microscopically negative margins, 5 (33%) had specimens that had less than 5% residual cancer cells, and 2 (13%) had specimens that had pathologic complete responses. The median overall survival of all patients was 21.7 months (95% CI, 15.7 to not reached) from registration.

Conclusions and relevance: The successful completion of this collaborative study demonstrates the feasibility of conducting quality-controlled trials for this disease stage in the multi-institutional setting. The data generated by this study and the logistical elements that facilitated the trial's completion are currently being used to develop cooperative group trials with the goal of improving outcomes for this subset of patients.

Trial registration: clinicaltrials.gov Identifier: NCT01821612.

Figure 1. CONSORT Diagram Including All 29 Patients Who Preregistered for the Trial

mFOLFIRINOX indicates modified treatment with 85 mg/m² of oxaliplatin, 180 mg/m² of irinotecan hydrochloride, 400 mg/m² of leucovorin calcium, and then 2400 mg/m² of 5-fluorouracil for 4 cycles; RECIST, Response Evaluation Criteria in Solid Tumors.

1. An additional patient had progression of disease determined by RECIST owing to isolated progression after mFOLFIRINOX treatment but, per protocol, proceeded to chemoradiation.

2. Treatment was halted for 1 patient after first cycle of postoperative chemotherapy.

Figure 2. Kaplan-Meier Survival Curves

A, The median overall survival of all 22 patients was 21.7 months (95% CI, 15.7 to not reached) from registration. B, Patients who underwent a pancreatectomy had a significantly higher 18-month overall survival rate than patients who did not (67% vs 43%; hazard ratio, 0.13 [95%CI, 0.03–0.48]; P = .001). C, Patients who underwent a pancreatectomy and had less than 5% cancer cells in their surgical specimen had a longer overall survival than patients who had 5% or more cancer cells (median overall survival not evaluable vs 21.7 months [95% CI, 15.9–25.2 months]; P = .01). Hash marks represent censored data.