Background: Relationships between normal tissue reactions and tumor response to radiotherapy have generally been investigated in retrospective studies, with the attendant difficulties such as different times of treatment and lack of standard scales for scoring of normal tissue reactions. This study analyses the correlation between normal tissue reactions and tumor response in a randomized trial of hyperfractionation versus conventional radiotherapy.
Materials and methods: EORTC trial 22791, which accrued 325 T2-3, N0-1 and M0 oropharynx squamous cell carcinomas, compared conventional radiotherapy (70 Gy in 7 weeks) to a hyperfractionated regime (two daily irradiations of 1.15 Gy up to 80.5 Gy in 7 weeks). Acute and late toxicities were assessed according to the RTOG/EORTC scales. The variables were tumor regression at treatment completion, tumor control for unlimited follow-up, severity of acute mucosal reactions (grade of the peak reaction, time to onset of grade 2 mucositis, or duration of acute reactions of grade 3) and severity of late effects in normal tissues (late ulceration, cervical edema secondary to lymphatic drainage damage, late muscular fibrosis, or late mucosal necrosis as consequential damage). Kendall's rank correlation was used to assess the levels of significance of the correlation. The 95% confidence interval (95% CI) was calculated for each correlation tau-b coefficient.
Results: Irradiation doses and overall treatment times were consistent between the treatment arms. The time to onset of patchy mucosal reactions was inversely correlated to tumor regression at treatment completion in the hyperfractionated arm (arm 2) but not in the conventional fractionation arm (arm 1). The other significant correlations were restricted to the following pairs of end-points and patient subgroups: acute mucositis of grade 3 versus tumor regression at the end of treatment in arm 2 for T2 tumors, late ulceration versus tumor control in arm 2 and acute mucosal reactions versus late effects in normal tissues in arm 2 for T3 tumors. By and large, the 95% CIs for Kendall's tau-b included zero and did not include 0.5, an indication that the comparison had sufficient power to detect a difference, if indeed one existed.
Conclusions: These results do not represent clinical evidence strong enough to demonstrate that individual differences in sensitivity influence the response of tumors to radiation.