Radiotherapy-related early morbidity in head and neck cancer: quantitative clinical radiobiology as deduced from the CHART trial

Background and purpose: Early reactions after radiotherapy for head and neck cancer may become the limiting factor in current attempts to intensify loco-regional therapy through altered fractionation or combination of radiotherapy with chemotherapy. The aim of the present study was to quantify the dependence of early reactions on the dose-fractionation used in radiotherapy and on patients' age and radiation field size. Patients and methods: The data analyzed are from the randomized controlled trial of CHART (continuous hyperfractionated accelerated radiotherapy) vs. conventional radiotherapy in head and neck cancer. The trial accrued 918 patients from March 1990 to April 1995. Several early morbidity items were evaluated weekly for the first 8 weeks after the start of treatment. Weekly evaluation was continued in patients with early reactions extending beyond 8 weeks. Linear regression was used to analyze the time with reactions in individual patients. Polychotomous ordinal response regression was used to analyze the peak grade of early reactions in individual patients. Results: The main findings of this analysis were as follows. (1) The incidence and peak prevalence of confluent mucositis was higher after CHART than after conventional radiotherapy. Therefore, the average time spent with confluent mucositis per patient treated was significantly longer after CHART than after conventional fractionation. (2) In patients who actually developed confluent mucositis, the average duration of this grade of reaction was not significantly different after CHART compared with conventional radiotherapy. (3) Confluent mucositis developed earlier after the start of treatment (2.9 vs. 4.9 weeks) but also started to improve sooner (5.4 vs. 7.5 weeks after the start of treatment) after CHART than after conventional radiotherapy. (4) The dose recovered per 1-day protraction of overall treatment time, D-prolif, was estimated at 0.80 Gy with 95% confidence limits 0.7 and 1. 1 Gy/day for human mucosa. (5) For human skin erythema, the estimate of D-prolif was 0.12 Gy/day with 95% confidence limits -0.12 and 0.22 Gy/day. (6) Highly significant relationships were found between the grade of morphological mucositis on one hand and dysphagia, pain on swallowing and prescribed analgesics on the other. Patients with confluent mucositis had fewer functional problems if this was confined to the larynx as compared with other subsites in the head and neck. (7) Although the incidence of confluent mucositis was higher in the oral cavity and oropharynx than in the hypopharynx and larynx, the radiobiological properties of mucosal reactions did not show significant variation among the various subsites within the head and neck region. (8) For a given dose and overall treatment time, a highly significant increase in incidence and severity of both mucositis and erythema was seen with increasing field size. Thus, a significant dose-volume or dose-area effect exists for both of these tissues. (9) Patients' age had no significant influence on the incidence and severity of mucositis or erythema. Conclusions: This study provides quantitative estimates of the dose-time and dose-volume relationships for human skin and normal mucosa in the head and neck region based on an analysis of data from 918 patients entered into a randomized-control led trial of altered dose fractionation in radiotherapy. (C) 2001 Elsevier Science Ireland Ltd. All rights reserved.