Purpose: A German multicenter randomized trial (ARO 95-6) compared hyperfractionated accelerated radiotherapy (RT) alone (to a total radiation dose of 77.6 Gy) with hyperfractionated accelerated radiochemotherapy (to 70.6 Gy) using concurrent mitomycin C and 5-fluorouracil. We analyzed the baseline patient characteristics and the influence of physician selection bias on treatment outcome for patients who were and were not enrolled in the randomized Phase III trial, with the therapies administered according to the trial protocols.
Methods and materials: Between February 1996 and May 2000 at Tübingen University, 42 on-study patients and 41 off-study patients with Stage III-IV nonmetastatic squamous cell carcinoma of the head and neck were treated. The median follow-up for patients at risk (living at last evaluation) was 44 months, with a minimal follow-up of 2 years.
Results: The 4-year rate of overall survival, disease-specific survival, and locoregional tumor control was 25%, 40%, and 54%, respectively, for all 83 patients. Among patients enrolled in the study, the 4-year rate of overall survival for those receiving accelerated hyperfractionated radiochemotherapy was 33%, and that for patients receiving accelerated hyperfractionated RT alone was 18% (p = 0.25); among off-study patients, the comparable rates were 48% and 0% (p = 0.004). The 4-year rate of disease-specific survival among on-study patients receiving radiochemotherapy and RT alone was 41% and 36%, respectively (p = 0.5); among off-study patients the respective rates were 58% and 0% (p = 0.2). The rate of 4-year locoregional tumor control associated with radiochemotherapy and RT, respectively, was 51% and 54% among on-study patients and 72% and 23% among off-study patients (p = 0.08).
Conclusion: Patients with advanced head-and-neck cancer who were entered into the randomized trial did not have statistically significantly different survival than patients treated according to the same protocol but outside the trial. Also, outside the trial, the physicians' selection bias in determining which patient received which treatment showed a much greater benefit from combined modality treatment than that found in the randomized trial.