Purpose: To assess the impact of treatment interruption on the potential gain in locoregional control obtained with accelerated radiotherapy (RT) compared with conventionally fractionated RT in patients with oropharyngeal carcinomas.
Methods and materials: 152 patients treated with radical RT for oropharyngeal carcinomas between 1979 and 1996 were retrospectively analyzed. According to the American Joint Committee on Cancer (AJCC) staging system, there were 6/30/43/73 stages III/III/IV. Sixty-one patients were treated with a conventional RT schedule (median dose 70 Gy in 35 fractions), and 91 patients with either of two 5/5.5-week accelerated RT schedules (median dose 69.6-69.9 Gy in 41 fractions). Discounting weekends, RT was interrupted for 2 consecutive days or more in 53 patients (median duration 11 days, range 2-97), including 67% of the patients in the conventional RT group and 13% in the accelerated RT group. Median follow-up for surviving patients was 55 months (range 23-230). The Cox proportional hazards model was used for the multivariate analysis of factors influencing locoregional control.
Results: In univariate analysis, factors associated with a significant decrease in locoregional control included WHO performance status > or =1, advanced AJCC stages (III and IV), conventional RT fractionation, overall treatment time > or =44 days (median), and RT interruption. In the multivariate analysis, when introduced into the model individually, the three significant therapeutic factors remained significant after adjustment for the forced clinical variables. However, when the three therapeutic factors were introduced together into the model, beside the AJCC stage (P = 0.017), only RT interruption remained a significant independent adverse prognostic factor (P = 0.026).
Conclusions: This multivariate analysis highlights the potential negative impact of treatment gaps on locoregional control in oropharyngeal carcinomas. This suggests that treatment interruption may be an even more important parameter than the type of RT schedule per se. Thus, when assessing the relative merit of two RT schedules, inclusion of the other therapeutic factors in a multivariate model is mandatory in order to avoid misinterpretation of the results.